Dacheng Tao

AAAI Conference 2026 Conference Paper

AD-FM: Multimodal LLMs for Anomaly Detection via Multi-Stage Reasoning and Fine-Grained Reward Optimization

• Jingyi Liao
• Yongyi Su
• Rong-Cheng Tu
• Zhao Jin
• Wenhao Sun
• Yiting Li
• Xun Xu
• Dacheng Tao

While Multimodal Large Language Models (MLLMs) demonstrate remarkable capabilities across diverse domains, their application to specialized anomaly detection (AD) remains constrained by domain adaptation challenges. Existing Group Relative Policy Optimization (GRPO) based approaches suffer from two critical limitations: inadequate training data utilization when models produce uniform responses, and insufficient supervision over reasoning processes that encourage immediate binary decisions without deliberative analysis. We propose a comprehensive framework addressing these limitations through two synergistic innovations. First, we introduce a multi-stage deliberative reasoning process that guides models from region identification to focused examination, generating diverse response patterns essential for GRPO optimization while enabling structured supervision over analytical workflows. Second, we develop a fine-grained reward mechanism incorporating classification accuracy and localization supervision, transforming binary feedback into continuous signals that distinguish genuine analytical insight from spurious correctness. Comprehensive evaluation across multiple industrial datasets shows that our method achieves superior accuracy by enabling general-purpose MLLMs to acquire fine-grained visual discrimination for detecting subtle manufacturing defects.

AAAI Conference 2026 Conference Paper

Bridging the Tokenizer Gap: Semantics and Distribution-aware Knowledge Transfer for Unbiased Cross-Tokenizer Distillation

• Huazheng Wang
• Yongcheng Jing
• Haifeng Sun
• Jingyu Wang
• Jianxin Liao
• Leszek Rutkowski
• Dacheng Tao

Cross-tokenizer knowledge distillation, where the teacher and student employ different tokenizers, is becoming increasingly prevalent, yet it poses underexplored challenges: existing methods fail to capture the rich knowledge encoded in teacher logits, as evidenced by the neglect of semantic information, inaccurate and biased logit alignment, and discarding distributional structure—ultimately leading to unfavorable distillation. To address these issues, we propose SeDi, a semantics and distribution-aware knowledge transfer framework tailored for cross-tokenizer distillation. To preserve factual knowledge, SeDi employs bipartite graph-based alignment at the tokenization level and a sliding window re-encoding strategy at the vocabulary level, enabling unbiased transfer of the teacher’s next-token predictions into the student’s vocabulary space. To further retain distributional information, we align the student’s entropy with that of the teacher by incorporating the student’s own logits during training, which helps to mitigate the exposure bias problem. Experiments on ten datasets across three task domains and five different teacher-student model pairs with varying vocabulary sizes demonstrate that SeDi delivers substantial improvements, with gains of up to 19.8%.

AAAI Conference 2026 Conference Paper

CogniTrust: Cognitive Memory-Driven Verifiable Supervision for Robust Hashing

• Yiyang Gu
• Bohan Wu
• Yifang Qin
• Jiaru Tang
• Rong-Cheng Tu
• Zhiping Xiao
• Taian Guo
• Junyu Luo

In this paper, we study the problem of robust multi-label hashing, where label noise hinders the learning of a reliable semantic structure from data. Many existing methods rely on heuristic sample selection or consistency-based training, but lack a unified mechanism to validate and refine supervision across structural and semantic levels. Inspired by cognitive theories of human memory, we propose a novel framework called CogniTrust that unifies verifiable supervision with a triadic memory model: a) In episodic memory, feature activations are decomposed into spatial patterns that support the assessment of structural evidence and the estimation of label reliability; b) Semantic memory keeps track of class-level prototypes from structurally attentive regions to estimate the semantic plausibility of labels; c) Reconstructive memory simulates memory recall through interpolation between images using a diffusion-based mixup process, which enriches the training signals for semantically uncertain regions. These components work together, allowing supervision to be refined through the joint consideration of spatial structure and semantic information. Extensive experiments on noisy hashing benchmarks demonstrate that CogniTrust consistently outperforms a range of state-of-the-art baselines. Our results show that cognitive memory mechanisms offer a principled basis for more reliable label denoising and robust hashing.

TMLR Journal 2026 Journal Article

Drawback of Enforcing Equivariance and its Compensation via the Lens of Expressive Power

• Yuzhu Chen
• Tian Qin
• Xinmei Tian
• Fengxiang He
• Dacheng Tao

Equivariant neural networks encode the intrinsic symmetry of data as an inductive bias, which has achieved impressive performance in wide domains. However, the understanding to their expressive power remains premature. Focusing on 2-layer ReLU networks, this paper investigates the impact of enforcing equivariance constraints on the expressive power. By examining the boundary hyperplanes and the channel vectors, we constructively demonstrate that enforcing equivariance constraints could undermine the expressive power. Naturally, this drawback can be compensated for by enlarging the model size -- we further prove upper bounds on the required enlargement for compensation. Surprisingly, we show that the enlarged neural architectures have reduced hypothesis space dimensionality, implying even better generalizability.

AAAI Conference 2026 Conference Paper

Free-Form Scene Editor: Enabling Multi-Round Object Manipulation Like in a 3D Engine

• Xincheng Shuai
• Zhenyuan Qin
• Henghui Ding
• Dacheng Tao

Recent advances in text-to-image (T2I) diffusion models have significantly improved semantic image editing, yet most methods fall short in performing 3D-aware object manipulation. In this work, we present FFSE, a 3D-aware autoregressive framework designed to enable intuitive, physically-consistent object editing directly on real-world images. Unlike previous approaches that either operate in image space or require slow and error-prone 3D reconstruction, FFSE models editing as a sequence of learned 3D transformations, allowing users to perform arbitrary manipulations, such as translation, scaling, and rotation, while preserving realistic background effects (e.g., shadows, reflections) and maintaining global scene consistency across multiple editing rounds. To support learning of multi-round 3D-aware object manipulation, we introduce 3DObjectEditor, a hybrid dataset constructed from simulated editing sequences across diverse objects and scenes, enabling effective training under multi-round and dynamic conditions. Extensive experiments show that the proposed FFSE significantly outperforms existing methods in both single-round and multi-round 3D-aware editing scenarios.

AAAI Conference 2026 Conference Paper

Lightning Fast Caching-based Parallel Denoising Prediction for Accelerating Talking Head Generation

• Jianzhi Long
• Wenhao Sun
• Rong-Cheng Tu
• Dacheng Tao

Diffusion-based talking head models generate high-quality, photorealistic videos but suffer from slow inference, limiting practical applications. Existing acceleration methods for gen- eral diffusion models fail to exploit the temporal and spatial redundancies unique to talking head generation. In this paper, we propose a task-specific framework addressing these inefficiencies through two key innovations. First, we introduce Lightning-fast Caching-based Parallel denoising prediction (LightningCP), caching static features to bypass most model layers in inference time. We also enable parallel prediction using cached features and estimated noisy latents as inputs, efficiently bypassing sequential sampling. Second, we propose Decoupled Foreground Attention (DFA) to further accelerate attention computations, exploiting the spatial decoupling in talking head videos to restrict attention to dynamic foreground regions. Additionally, we remove reference features in certain layers to bring extra speedup. Extensive experiments demonstrate that our framework significantly improves inference speed while preserving video quality.

AAAI Conference 2026 Conference Paper

SRD: Reinforcement-Learned Semantic Perturbation for Backdoor Defense in VLMs

• Shuhan Xu
• Siyuan Liang
• Hongling Zheng
• Aishan Liu
• Xinbiao Wang
• Yong Luo
• Fu Lin
• Leszek Rutkowski

Visual language models (VLMs) have made significant progress in image captioning tasks, yet recent studies have found they are vulnerable to backdoor attacks. Attackers can inject undetectable perturbations into the data during inference, triggering abnormal behavior and generating malicious captions. These attacks are particularly challenging to detect and defend against due to the stealthiness and cross-modal propagation of the trigger signals. In this paper, we identify two key vulnerabilities by analyzing existing attack patterns: (1) the model exhibits abnormal attention concentration on certain regions of the input image, and (2) backdoor attacks often induce semantic drift and sentence incoherence. Based on these insights, we propose Semantic Reward Defense (SRD), a reinforcement learning framework that mitigates backdoor behavior without requiring any prior knowledge of trigger patterns. SRD learns to apply discrete perturbations to sensitive contextual regions of image inputs via a deep Q-network policy, aiming to confuse attention and disrupt the activation of malicious paths. To guide policy optimization, we design a reward signal named semantic fidelity score, which jointly assesses the semantic consistency and linguistic fluency of the generated captions, encouraging the agent to achieve a robust yet faithful output. SRD offers a trigger-agnostic, policy-interpretable defense paradigm that effectively mitigates local (TrojVLM) and global (Shadowcast) backdoor attacks, reducing ASR to 3.6% and 5.6% respectively, with less than 15% average CIDEr drop on the clean inputs.

TMLR Journal 2026 Journal Article

Subspace based Federated Unlearning

• Guanghao Li
• Li Shen
• Yan Sun
• Yue Hu
• Han Hu
• Dacheng Tao

Federated learning (FL) enables collaborative machine learning among multiple clients while preserving user data privacy by preventing the exchange of local data. However, when users request to leave the FL system, the trained FL model may still retain information about their contributions. To comply with the right to be forgotten, federated unlearning has been proposed, which aims to remove a designated client's influence from the FL model. Existing federated unlearning methods typically rely on storing historical parameter updates, which may be impractical in resource-constrained FL settings. In this paper, we propose a Subspace-based Federated Unlearning method (SFU) that addresses this challenge without requiring additional storage. SFU updates the model via gradient ascent constrained within a subspace, specifically the orthogonal complement of the gradient descent directions derived from the remaining clients. By projecting the ascending gradient of the target client onto this subspace, SFU can mitigate the contribution of the target client while maintaining model performance on the remaining clients. SFU is communication-efficient, requiring only one round of local training per client to transmit gradient information to the server for model updates. Extensive empirical evaluations on multiple datasets demonstrate that SFU achieves competitive unlearning performance while preserving model utility. Compared to representative baseline methods, SFU consistently shows promising results under various experimental settings.

AAAI Conference 2026 Conference Paper

Time-Frequency Token Advantage Clipping for Training Efficient Large Reasoning Model

• Rong Bao
• Bo Wang
• Xiao Wang
• Hongyu Li
• Rui Zheng
• Leszek Rutkowski
• Qi Zhang
• Liang Ding

Long Chain-of-Thought (CoT) reasoning enhances large reasoning models' performance but suffers from severe inefficiencies, as models often overthink simple problems or underthink complex ones. Current sequence-level optimizations, like length penalties, are too coarse-grained to distinguish core logic from verbose language, precluding the necessary token-level control for efficient reasoning CoT. To overcome these limitations, we introduce Time-Frequency token Advantage Clipping (TFAC), a novel training framework designed to build efficient large reasoning models via token-level interventions. Specifically, TFAC functions along two dimensions: 1) The Frequency Dimension: It discourages inefficient loops and encourages deeper exploration by dynamically reducing the advantage scores of high-entropy tokens that are repeatedly generated within a single reasoning path. 2) The Time Dimension: It reduces excessive overthinking of the system by establishing a historical baseline for the occurrence count of each critical token in previously successful trajectories, and clipping the advantages of tokens that exceed this baseline during training. Crucially, to preserve the model's exploratory capabilities on novel problems, this suppression mechanism is automatically disabled when no historical record of success is available. Experiments conducted on the Deepseek-Distill-32B and Qwen3-8B models show that TFAC outperforms leading baseline methods, improving performance by 2.3 and 3.1 percentage points, respectively, while simultaneously reducing inference costs by 35% and 28% in scenarios where correct answers are generated. These results validate the significant efficacy of TFAC in training large reasoning models that are both powerful and highly efficient.

ICLR Conference 2025 Conference Paper

A Statistical Approach for Controlled Training Data Detection

• Zirui Hu
• Yingjie Wang
• Zheng Zhang
• Hong Chen
• Dacheng Tao

Detecting training data for large language models (LLMs) is receiving growing attention, especially in applications requiring high reliability. While numerous efforts have been made to address this issue, they typically focus on accuracy without ensuring controllable results. To fill this gap, we propose **K**nockoff Inference-based **T**raining data **D**etector (KTD), a novel method that achieves rigorous false discovery rate (FDR) control in training data detection. Specifically, KTD generates synthetic knockoff samples that seamlessly replace original data points without compromising contextual integrity. A novel knockoff statistic, which incorporates multiple knockoff draws, is then calculated to ensure FDR control while maintaining high power. Our theoretical analysis demonstrates KTD's asymptotic optimality in terms of FDR control and power. Empirical experiments on real-world datasets such as WikiMIA, XSum and Real Time BBC News further validate KTD's superior performance compared to existing methods.

ICLR Conference 2025 Conference Paper

A Theoretical Perspective: How to Prevent Model Collapse in Self-consuming Training Loops

• Shi Fu
• Yingjie Wang
• Yuzhu Chen
• Xinmei Tian 0001
• Dacheng Tao

High-quality data is essential for training large generative models, yet the vast reservoir of real data available online has become nearly depleted. Consequently, models increasingly generate their own data for further training, forming Self-consuming Training Loops (STLs). However, the empirical results have been strikingly inconsistent: some models degrade or even collapse, while others successfully avoid these failures, leaving a significant gap in theoretical understanding to explain this discrepancy. This paper introduces the intriguing notion of *recursive stability* and presents the first theoretical generalization analysis, revealing how both model architecture and the proportion between real and synthetic data influence the success of STLs. We further extend this analysis to transformers in in-context learning, showing that even a constant-sized proportion of real data ensures convergence, while also providing insights into optimal synthetic data sizing.

NeurIPS Conference 2025 Conference Paper

Ada-R1: Hybrid-CoT via Bi-Level Adaptive Reasoning Optimization

• Haotian Luo
• Haiying He
• Yibo Wang
• Jinluan Yang
• Rui Liu
• Naiqiang Tan
• Xiaochun Cao
• Dacheng Tao

Recently, long-thought reasoning models achieve strong performance on complex reasoning tasks, but often incur substantial inference overhead, making efficiency a critical concern. Our empirical analysis reveals that the benefit of using Long-CoT varies across problems: while some problems require elaborate reasoning, others show no improvement—or even degraded accuracy. This motivates adaptive reasoning strategies that tailor reasoning depth to the input. However, prior work primarily reduces redundancy within long reasoning paths, limiting exploration of more efficient strategies beyond the Long-CoT paradigm. To address this, we propose a novel two-stage framework for adaptive and efficient reasoning. First, we construct a hybrid reasoning model by merging long and short CoT models to enable diverse reasoning styles. Second, we apply bi-level preference training to guide the model to select suitable reasoning styles (group-level), and prefer concise and correct reasoning within each style group (instance-level). Experiments demonstrate that our method significantly reduces inference costs compared to other baseline approaches, while maintaining performance. Notably, on five mathematical datasets, the average length of reasoning is reduced by more than 50\%, highlighting the potential of adaptive strategies to optimize reasoning efficiency in large language models.

NeurIPS Conference 2025 Conference Paper

Adaptive Defense against Harmful Fine-Tuning for Large Language Models via Bayesian Data Scheduler

• Zixuan Hu
• Li Shen
• Zhenyi Wang
• Yongxian Wei
• Dacheng Tao

Harmful fine-tuning poses critical safety risks to fine-tuning-as-a-service for large language models. Existing defense strategies preemptively build robustness via attack simulation but suffer from fundamental limitations: (i) the infeasibility of extending attack simulations beyond bounded threat models due to the inherent difficulty of anticipating unknown attacks, and (ii) limited adaptability to varying attack settings, as simulation fails to capture their variability and complexity. To address these challenges, we propose Bayesian Data Scheduler (BDS), an adaptive tuning-stage defense strategy with no need for attack simulation. BDS formulates harmful fine-tuning defense as a Bayesian inference problem, learning the posterior distribution of each data point's safety attribute, conditioned on the fine-tuning and alignment datasets. The fine-tuning process is then constrained by weighting data with their safety attributes sampled from the posterior, thus mitigating the influence of harmful data. By leveraging the post hoc nature of Bayesian inference, the posterior is conditioned on the fine-tuning dataset, enabling BDS to tailor its defense to the specific dataset, thereby achieving adaptive defense. Furthermore, we introduce a neural scheduler based on amortized Bayesian learning, enabling efficient transfer to new data without retraining. Comprehensive results across diverse attack and defense settings demonstrate the state-of-the-art performance of our approach. Code is available at https: //github. com/Egg-Hu/Bayesian-Data-Scheduler.

NeurIPS Conference 2025 Conference Paper

AdaVideoRAG: Omni-Contextual Adaptive Retrieval-Augmented Efficient Long Video Understanding

• Xue zhucun
• Jiangning Zhang
• Xie Xurong
• Yuxuan Cai
• Yong Liu
• Xiangtai Li
• Dacheng Tao

Multimodal Large Language Models (MLLMs) have demonstrated excellent performance in video understanding but suffer from degraded effectiveness when processing long videos due to fixed-length contexts and weaknesses in modeling long-term dependencies. Retrieval-Augmented Generation (RAG) technology can mitigate these limitations through dynamic knowledge expansion, but existing RAG schemes for video understanding employ fixed retrieval paradigms that use uniform structures regardless of input query difficulty. This introduces redundant computational overhead and latency ( e. g. , complex graph traversal operations) for simple queries ( e. g. , frame-level object recognition) while potentially causing critical information loss due to insufficient retrieval granularity for multi-hop reasoning. Such single-step retrieval mechanisms severely constrain the model's balance between resource efficiency and cognitive depth. To address this, we first propose a novel AdaVideoRAG framework for long-video understanding, which uses a lightweight intent classifier to dynamically and adaptively allocate appropriate retrieval schemes, ranging from the simplest to the most sophisticated, for different video understanding tasks based on query complexity. We introduce an Omni-Knowledge Indexing module to extract valuable information from multi-modal signals for context modeling and build corresponding databases, i. e. , a text base from clip captions, ASR, and OCR; a visual base; and a graph for deep semantic understanding. This enables hierarchical knowledge access, integration, and generation from naive retrieval to graph retrieval, achieving an optimal balance between resource consumption and video understanding capabilities. Finally, we construct the HiVU benchmark for deep understanding evaluation. Extensive experiments show that our framework enhances the overall efficiency and accuracy of Video-QA for long videos and can be seamlessly integrated with existing MLLMs via lightweight API calls, establishing a new paradigm for adaptive retrieval augmentation in video analysis.

NeurIPS Conference 2025 Conference Paper

AiDE-Q: Synthetic Labeled Datasets Can Enhance Learning Models for Quantum Property Estimation

• Xinbiao Wang
• Yuxuan Du
• Zihan Lou
• Yang Qian
• Kaining Zhang
• Yong Luo
• Bo Du
• Dacheng Tao

Quantum many-body problems are central to various scientific disciplines, yet their ground-state properties are intrinsically challenging to estimate. Recent advances in deep learning (DL) offer potential solutions in this field, complementing prior purely classical and quantum approaches. However, existing DL-based models typically assume access to a large-scale and noiseless labeled dataset collected by infinite sampling. This idealization raises fundamental concerns about their practical utility, especially given the limited availability of quantum hardware in the near term. To unleash the power of these DL-based models, we propose AiDE-Q (\underline{a}utomat\underline{i}c \underline{d}ata \underline{e}ngine for \underline{q}uantum property estimation), an effective framework that addresses this challenge by iteratively generating high-quality synthetic labeled datasets. Specifically, AiDE-Q utilizes a confidence-check method to assess the quality of synthetic labels and continuously improves the employed DL models with the identified high-quality synthetic dataset. To verify the effectiveness of AiDE-Q, we conduct extensive numerical simulations on a diverse set of quantum many-body and molecular systems, with up to 50 qubits. The results show that AiDE-Q enhances prediction performance for various reference learning models, with improvements of up to $14. 2\\%$. Moreover, we exhibit that a basic supervised learning model integrated with AiDE-Q outperforms advanced reference models, highlighting the importance of a synthetic dataset. Our work paves the way for more efficient and practical applications of DL for quantum property estimation.

NeurIPS Conference 2025 Conference Paper

Analytic Energy-Guided Policy Optimization for Offline Reinforcement Learning

• Jifeng Hu
• Sili Huang
• Zhejian Yang
• Shengchao Hu
• Li Shen
• Hechang Chen
• Lichao Sun
• Yi Chang

Conditional decision generation with diffusion models has shown powerful competitiveness in reinforcement learning (RL). Recent studies reveal the relation between energy-function-guidance diffusion models and constrained RL problems. The main challenge lies in estimating the intermediate energy, which is intractable due to the log-expectation formulation during the generation process. To address this issue, we propose the Analytic Energy-guided Policy Optimization (AEPO). Specifically, we first provide a theoretical analysis and the closed-form solution of the intermediate guidance when the diffusion model obeys the conditional Gaussian transformation. Then, we analyze the posterior Gaussian distribution in the log-expectation formulation and obtain the target estimation of the log-expectation under mild assumptions. Finally, we train an intermediate energy neural network to approach the target estimation of log-expectation formulation. We apply our method in 30+ offline RL tasks to demonstrate the effectiveness of our method. Extensive experiments illustrate that our method surpasses numerous representative baselines in D4RL offline reinforcement learning benchmarks.

TMLR Journal 2025 Journal Article

Are Large Language Models Really Robust to Word-Level Perturbations?

• Haoyu Wang
• Guozheng Ma
• Cong Yu
• Ning Gui
• Linrui Zhang
• Zhiqi Huang
• Suwei Ma
• Yongzhe Chang

The swift advancement in the scales and capabilities of Large Language Models (LLMs) positions them as promising tools for a variety of downstream tasks. In addition to the pursuit of better performance and the avoidance of violent feedback on a certain prompt, to ensure the responsibility of the LLMs, much attention is drawn to the robustness of LLMs. However, existing evaluation methods mostly rely on traditional question answering datasets with predefined supervised labels, potentially ignoring the superior generation capabilities of contemporary LLMs. To investigate the robustness of LLMs while using their generation ability, we propose a novel rational evaluation pipeline that leverages reward models as diagnostic tools to evaluate the long conversation generated from more challenging open questions by LLMs, which we refer to as the Reward Model for Reasonable Robustness Evaluation (TREvaL). Longer conversations manifest the comprehensive grasp of language models in terms of their proficiency in understanding questions, a capability not entirely encompassed by individual words or letters.Our extensive empirical experiments demonstrate that TREvaL provides an identification for the lack of robustness of nowadays LLMs.Notably, we are surprised to discover that robustness tends to decrease as fine-tuning (SFT and RLHF) is conducted, calling for more attention on the robustness during alignment process.

ICML Conference 2025 Conference Paper

AsymRnR: Video Diffusion Transformers Acceleration with Asymmetric Reduction and Restoration

• Wenhao Sun
• Rong-Cheng Tu
• Jingyi Liao
• Zhao Jin
• Dacheng Tao

Diffusion Transformers (DiTs) have proven effective in generating high-quality videos but are hindered by high computational costs. Existing video diffusion sampling acceleration methods often rely on costly fine-tuning or exhibit limited generalization capabilities. We propose Asymmetric Reduction and Restoration ( AsymRnR ), a training-free and model-agnostic method to accelerate video DiTs. It builds on the observation that redundancies of feature tokens in DiTs vary significantly across different model blocks, denoising steps, and feature types. Our AsymRnR asymmetrically reduces redundant tokens in the attention operation, achieving acceleration with negligible degradation in output quality and, in some cases, even improving it. We also tailored a reduction schedule to distribute the reduction across components adaptively. To further accelerate this process, we introduce a matching cache for more efficient reduction. Backed by theoretical foundations and extensive experimental validation, AsymRnR integrates into state-of-the-art video DiTs and offers substantial speedup.

ICML Conference 2025 Conference Paper

Be Confident: Uncovering Overfitting in MLLM Multi-Task Tuning

• Wenke Huang 0003
• Jian Liang 0003
• Guancheng Wan
• Didi Zhu
• He Li 0054
• Jiawei Shao
• Mang Ye
• Bo Du 0001

Fine-tuning Multimodal Large Language Models (MLLMs) in multi-task learning scenarios has emerged as an effective strategy for achieving cross-domain specialization. However, multi-task fine-tuning frequently induces performance degradation on open-response datasets. We posit that free-form answer generation primarily depends on language priors, and strengthening the integration of visual behavioral cues is critical for enhancing prediction robustness. In this work, we propose Noise Resilient Confidence Alignment to address the challenge of open-response overfitting during multi-task fine-tuning. Our approach prioritizes maintaining consistent prediction patterns in MLLMs across varying visual input qualities. To achieve this, we employ Gaussian perturbations to synthesize distorted visual inputs and enforce token prediction confidence alignment towards the normal visual branch. By explicitly linking confidence calibration to visual robustness, this method reduces over-reliance on language priors. We conduct extensive empirical evaluations across diverse multi-task downstream settings via popular MLLM architectures. The comprehensive experiment demonstrates the effectiveness of our method, showcasing its ability to alleviate open-response overfitting while maintaining satisfying multi-task fine-tuning performance.

ICLR Conference 2025 Conference Paper

Combatting Dimensional Collapse in LLM Pre-Training Data via Submodular File Selection

• Ziqing Fan
• Siyuan Du
• Shengchao Hu
• Pingjie Wang
• Li Shen 0008
• Ya Zhang 0002
• Dacheng Tao
• Yanfeng Wang 0001

Selecting high-quality pre-training data for large language models (LLMs) is crucial for enhancing their overall performance under limited computation budget, improving both training and sample efficiency. Recent advancements in file selection primarily rely on using an existing or trained proxy model to assess the similarity of samples to a target domain, such as high quality sources BookCorpus and Wikipedia. However, upon revisiting these methods, the domain-similarity selection criteria demonstrates a diversity dilemma, i.e. dimensional collapse in the feature space, improving performance on the domain-related tasks but causing severe degradation on generic performance.To prevent collapse and enhance diversity, we propose a DiverSified File selection algorithm (DiSF), which selects the most decorrelated text files in the feature space. We approach this with a classical greedy algorithm to achieve more uniform eigenvalues in the feature covariance matrix of the selected texts, analyzing its approximation to the optimal solution under a formulation of $\gamma$-weakly submodular optimization problem. Empirically, we establish a benchmark and conduct extensive experiments on the TinyLlama architecture with models from 120M to 1.1B parameters. Evaluating across nine tasks from the Harness framework, DiSF demonstrates a significant improvement on overall performance. Specifically, DiSF saves 98.5\% of 590M training files in SlimPajama, outperforming the full-data pre-training within a 50B training budget, and achieving about 1.5x training efficiency and 5x data efficiency. Source code is available at: https://github.com/MediaBrain-SJTU/DiSF.git.

NeurIPS Conference 2025 Conference Paper

Consistent Paths Lead to Truth: Self-Rewarding Reinforcement Learning for LLM Reasoning

• Kongcheng Zhang
• QI YAO
• Shunyu Liu
• Yingjie Wang
• Baisheng Lai
• Jieping Ye
• Mingli Song
• Dacheng Tao

Recent advances of Reinforcement Learning (RL) have highlighted its potential in complex reasoning tasks, yet effective training often relies on external supervision, which limits the broader applicability. In this work, we propose a novel self-rewarding reinforcement learning framework to enhance Large Language Model (LLM) reasoning by leveraging the consistency of intermediate reasoning states across different reasoning trajectories. Our key insight is that correct responses often exhibit consistent trajectory patterns in terms of model likelihood: their intermediate reasoning states tend to converge toward their own final answers ( high consistency ) with minimal deviation toward other candidates ( low volatility ). Inspired by this observation, we introduce CoVo, an intrinsic reward mechanism that integrates Co nsistency and Vo latility via a robust vector-space aggregation strategy, complemented by a curiosity bonus to promote diverse exploration. CoVo enables LLMs to perform RL in a self-rewarding manner, offering a scalable pathway for learning to reason without external supervision. Extensive experiments on diverse reasoning benchmarks show that CoVo achieves performance comparable to or even surpassing supervised RL. Our code is available at https: //github. com/sastpg/CoVo.

NeurIPS Conference 2025 Conference Paper

D$^2$GS: Dense Depth Regularization for LiDAR-free Urban Scene Reconstruction

• Kejing Xia
• Jidong Jia
• Ke Jin
• Yucai BAI
• Li Sun
• Dacheng Tao
• Youjian Zhang

Recently, Gaussian Splatting (GS) has shown great potential for urban scene reconstruction in the field of autonomous driving. However, current urban scene reconstruction methods often depend on multimodal sensors as inputs, $\textit{i. e. }$ LiDAR and images. Though the geometry prior provided by LiDAR point clouds can largely mitigate ill-posedness in reconstruction, acquiring such accurate LiDAR data is still challenging in practice: i) precise spatiotemporal calibration between LiDAR and other sensors is required, as they may not capture data simultaneously; ii) reprojection errors arise from spatial misalignment when LiDAR and cameras are mounted at different locations. To avoid the difficulty of acquiring accurate LiDAR depth, we propose ${D}^2GS$, a LiDAR-free urban scene reconstruction framework. In this work, we obtain geometry priors that are as effective as LiDAR while being denser and more accurate. $\textbf{First}$, we initialize a dense point cloud by back-projecting multi-view metric depth predictions. This point cloud is then optimized by a Progressive Pruning strategy to improve the global consistency. $\textbf{Second}$, we jointly refine Gaussian geometry and predicted dense metric depth via a Depth Enhancer. Specifically, we leverage diffusion priors from a depth foundation model to enhance the depth maps rendered by Gaussians. In turn, the enhanced depths provide stronger geometric constraints during Gaussian training. $\textbf{Finally}$, we improve the accuracy of ground geometry by constraining the shape and normal attributes of Gaussians within road regions. Extensive experiments on the Waymo dataset demonstrate that our method consistently outperforms state-of-the-art methods, producing more accurate geometry even when compared with those using ground-truth LiDAR data.

ICML Conference 2025 Conference Paper

Decision Mixer: Integrating Long-term and Local Dependencies via Dynamic Token Selection for Decision-Making

• Hongling Zheng
• Li Shen 0008
• Yong Luo 0002
• Deheng Ye
• Bo Du 0001
• Jialie Shen 0001
• Dacheng Tao

The Conditional Sequence Modeling (CSM) paradigm, benefiting from the transformer’s powerful distribution modeling capabilities, has demonstrated considerable promise in offline Reinforcement Learning (RL) tasks. Depending on the task’s nature, it is crucial to carefully balance the interplay between inherent local features and long-term dependencies in Markov decision trajectories to mitigate potential performance degradation and unnecessary computational overhead. In this paper, we propose Decision Mixer (DM), which addresses the conflict between features of different scales in the modeling process from the perspective of dynamic integration. Drawing inspiration from conditional computation, we design a plug-and-play dynamic token selection mechanism to ensure the model can effectively allocate attention to different features based on task characteristics. Additionally, we employ an auxiliary predictor to alleviate the short-sightedness issue in the autoregressive sampling process. DM achieves state-of-the-art performance on various standard RL benchmarks while requiring significantly fewer computational resources, offering a viable solution for building efficient and scalable RL foundation models. Code is available at here.

NeurIPS Conference 2025 Conference Paper

Detoxifying Large Language Models via Autoregressive Reward Guided Representation Editing

• Yisong Xiao
• Aishan Liu
• Siyuan Liang
• Zonghao Ying
• Xianglong Liu
• Dacheng Tao

Large Language Models (LLMs) have demonstrated impressive performance across various tasks, yet they remain vulnerable to generating toxic content, necessitating detoxification strategies to ensure safe and responsible deployment. Test-time detoxification methods, which typically introduce static or dynamic interventions into LLM representations, offer a promising solution due to their flexibility and minimal invasiveness. However, current approaches often suffer from imprecise interventions, primarily due to their insufficient exploration of the transition space between toxic and non-toxic outputs. To address this challenge, we propose \textsc{A}utoregressive \textsc{R}eward \textsc{G}uided \textsc{R}epresentation \textsc{E}diting (ARGRE), a novel test-time detoxification framework that explicitly models toxicity transitions within the latent representation space, enabling stable and precise reward-guided editing. ARGRE identifies non-toxic semantic directions and interpolates between toxic and non-toxic representations to reveal fine-grained transition trajectories. These trajectories transform sparse toxicity annotations into dense training signals, enabling the construction of an autoregressive reward model that delivers stable and precise editing guidance. At inference, the reward model guides an adaptive two-step editing process to obtain detoxified representations: it first performs directional steering based on expected reward gaps to shift representations toward non-toxic regions, followed by lightweight gradient-based refinements. Extensive experiments across 8 widely used LLMs show that ARGRE significantly outperforms leading baselines in effectiveness (-62. 21\% toxicity) and efficiency (-47. 58\% inference time), while preserving the core capabilities of the original model with minimal degradation. Our code is available at the \href{https: //anonymous. 4open. science/r/ARGRE-6291}{anonymous website}.

AAAI Conference 2025 Conference Paper

Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models

• Wenbin Wang
• Liang Ding
• Minyan Zeng
• Xiabin Zhou
• Li Shen
• Yong Luo
• Wei Yu
• Dacheng Tao

Multimodal large language models (MLLMs) have experienced significant advancements recently, but still struggle to recognize and interpret intricate details in high-resolution (HR) images effectively. While state-of-the-art (SOTA) MLLMs claim to process images at 4K resolution, existing MLLM benchmarks only support up to 2K, leaving the capabilities of SOTA models on true HR images largely untested. Furthermore, existing methods for enhancing HR image perception in MLLMs rely on computationally expensive visual instruction tuning. To address these limitations, we introduce HR-Bench, the first deliberately designed benchmark to rigorously evaluate MLLM performance on 4K & 8K images. Through extensive experiments, we demonstrate that while downsampling HR images leads to vision information loss, leveraging complementary modalities, e.g., text, can effectively compensate for this loss. Building upon this insight, we propose Divide, Conquer and Combine, a novel training-free framework for enhancing MLLM perception of HR images. Our method follows a three-staged approach: 1) Divide: recursively partitioning the HR image into patches and merging similar patches to minimize computational overhead, 2) Conquer: leveraging the MLLM to generate accurate textual descriptions for each image patch, and 3) Combine: utilizing the generated text descriptions to enhance the MLLM's understanding of the overall HR image. Extensive experiments show that: 1) the SOTA MLLM achieves 63% accuracy, which is markedly lower than the 87% accuracy achieved by humans on HR-Bench; 2) our method brings consistent and significant improvements (a relative increase of +6% on HR-Bench and +8% on general multimodal benchmarks).

ICLR Conference 2025 Conference Paper

Dynamic Neural Fortresses: An Adaptive Shield for Model Extraction Defense

• Siyu Luan
• Zhenyi Wang 0001
• Li Shen 0008
• Zonghua Gu 0001
• Chao Wu
• Dacheng Tao

Model extraction aims to acquire a pre-trained black-box model concealed behind a black-box API. Existing defense strategies against model extraction primarily concentrate on preventing the unauthorized extraction of API functionality. However, two significant challenges still need to be solved: (i) Neural network architecture of the API constitutes a form of intellectual property that also requires protection; (ii) The current practice of allocating the same network architecture to both attack and benign queries results in substantial resource wastage. To address these challenges, we propose a novel \textit{Dynamic Neural Fortresses} (DNF) defense method, employing a dynamic Early-Exit neural network, deviating from the conventional fixed architecture. Firstly, we facilitate the random exit of attack queries from the network at earlier layers. This strategic exit point selection significantly reduces the computational cost for attack queries. Furthermore, the random exit of attack queries from earlier layers introduces increased uncertainty for attackers attempting to discern the exact architecture, thereby enhancing architectural protection. On the contrary, we aim to facilitate benign queries to exit at later layers, preserving model utility, as these layers typically yield meaningful information. Extensive experiments on defending against various model extraction scenarios and datasets demonstrate the effectiveness of DNF, achieving a notable 2$\times$ improvement in efficiency and an impressive reduction of up to 12\% in clone model accuracy compared to SOTA defense methods. Additionally, DNF provides strong protection against neural architecture theft, effectively safeguarding network architecture from being stolen.

NeurIPS Conference 2025 Conference Paper

Effective Policy Learning for Multi-Agent Online Coordination Beyond Submodular Objectives

• Qixin Zhang
• Yan Sun
• Can Jin
• Xikun Zhang
• Yao Shu
• Puning Zhao
• Li Shen
• Dacheng Tao

In this paper, we present two effective policy learning algorithms for multi-agent online coordination(MA-OC) problem. The first one, **MA-SPL**, not only can achieve the optimal $(1-\frac{c}{e})$-approximation guarantee for the MA-OC problem with submodular objectives but also can handle the unexplored $\alpha$-weakly DR-submodular and $(\gamma, \beta)$-weakly submodular scenarios, where $c$ is the curvature of the investigated submodular functions, $\alpha$ denotes the diminishing-return(DR) ratio and the tuple$(\gamma, \beta)$ represents the submodularity ratios. Subsequently, in order to reduce the reliance on the unknown parameters $\alpha, \gamma, \beta$ inherent in the **MA-SPL** algorithm, we then introduce the second online algorithm named **MA-MPL**. This **MA-MPL** algorithm is entirely *parameter-free* and simultaneously can maintain the same approximation ratio as the first **MA-SPL** algorithm. The core of our **MA-SPL** and **MA-MPL** algorithms is a novel continuous-relaxation technique term as policy-based continuous extension. Compared with the well-established multi-linear extension, a notable advantage of this new policy-based continuous extension is its ability to provide a lossless rounding scheme for any set function, thereby enabling us to tackle the challenging weakly submodular objective functions. Finally, extensive simulations are conducted to demonstrate the effectiveness of our proposed algorithms.

ICLR Conference 2025 Conference Paper

Energy-based Backdoor Defense Against Federated Graph Learning

• Guancheng Wan
• Zitong Shi
• Wenke Huang 0003
• Guibin Zhang
• Dacheng Tao
• Mang Ye

Federated Graph Learning is rapidly evolving as a privacy-preserving collaborative approach. However, backdoor attacks are increasingly undermining federated systems by injecting carefully designed triggers that lead to the model making incorrect predictions. Trigger structures and injection locations in Federated Graph Learning are more diverse, making traditional federated defense methods less effective. In our work, we propose an effective Federated Graph Backdoor Defense using Topological Graph Energy (FedTGE). At the local client level, it injects distribution knowledge into the local model, assigning low energy to benign samples and high energy to the constructed malicious substitutes, and selects benign clients through clustering. At the global server level, the energy elements uploaded by each client are treated as new nodes to construct a global energy graph for energy propagation, making the selected clients' energy elements more similar and further adjusting the aggregation weights. Our method can handle high data heterogeneity, does not require a validation dataset, and is effective under both small and large malicious proportions. Extensive results on various settings of federated graph scenarios under backdoor attacks validate the effectiveness of this approach.

ICLR Conference 2025 Conference Paper

FreDF: Learning to Forecast in the Frequency Domain

• Hao Wang 0049
• Lichen Pan
• Yuan Shen
• Zhichao Chen 0001
• Degui Yang
• Yifei Yang
• Sen Zhang 0006
• Xinggao Liu

Time series modeling presents unique challenges due to autocorrelation in both historical data and future sequences. While current research predominantly addresses autocorrelation within historical data, the correlations among future labels are often overlooked. Specifically, modern forecasting models primarily adhere to the Direct Forecast (DF) paradigm, generating multi-step forecasts independently and disregarding label correlations over time. In this work, we demonstrate that the learning objective of DF is biased in the presence of label correlation. To address this issue, we propose the Frequency-enhanced Direct Forecast (FreDF), which mitigates label correlation by learning to forecast in the frequency domain, thereby reducing estimation bias. Our experiments show that FreDF significantly outperforms existing state-of-the-art methods and is compatible with a variety of forecast models. Code is available at https://github.com/Master-PLC/FreDF.

JMLR Journal 2025 Journal Article

FusionBench: A Unified Library and Comprehensive Benchmark for Deep Model Fusion

• Anke Tang
• Li Shen
• Yong Luo
• Enneng Yang
• Han Hu
• Lefei Zhang
• Bo Du
• Dacheng Tao

Deep model fusion is an emerging technique that unifies the predictions or parameters of several deep neural networks into a single better-performing model in a cost-effective and data-efficient manner. Although a variety of deep model fusion techniques have been introduced, their evaluations tend to be inconsistent and often inadequate to validate their effectiveness and robustness. We present FusionBench, the first benchmark and a unified library designed specifically for deep model fusion. Our benchmark consists of multiple tasks, each with different settings of models and datasets. This variety allows us to compare fusion methods across different scenarios and model scales. Additionally, FusionBench serves as a unified library for easy implementation and testing of new fusion techniques. FusionBench is open source and actively maintained, with community contributions encouraged. [abs] [ pdf ][ bib ] [ code ] &copy JMLR 2025. ( edit, beta )

IJCAI Conference 2025 Conference Paper

Human-Imperceptible, Machine-Recognizable Images

• Fusheng Hao
• Fengxiang He
• Yikai Wang
• Fuxiang Wu
• Jing Zhang
• Dacheng Tao
• Jun Cheng

Massive human-related data is collected to train neural networks for computer vision tasks. A major conflict is exposed relating to software engineers between better developing AI systems and distancing from the sensitive training data. To reconcile this conflict, the paper proposes an efficient privacy-preserving learning paradigm, where images are encrypted to become ``human-imperceptible, machine-recognizable'' via one of the two encryption strategies: (1) random shuffling equally-sized patches and (2) mixing-up sub-patches. Then, minimal adaptations are made to vision transformer to enable it to learn on the encrypted images for vision tasks, including image classification and object detection. Extensive experiments on ImageNet and COCO show that the proposed paradigm achieves comparable accuracy with the competitive methods. Decrypting the encrypted images requires solving an NP-hard jigsaw puzzle or ill-posed inverse problem, which is empirically shown intractable to be recovered by various attackers, including the powerful vision transformer-based attacker. We thus show that the proposed paradigm can ensure the encrypted images have become human-imperceptible while preserving machine-recognizable information.

ICML Conference 2025 Conference Paper

ICLShield: Exploring and Mitigating In-Context Learning Backdoor Attacks

• Zhiyao Ren
• Siyuan Liang 0004
• Aishan Liu
• Dacheng Tao

In-context learning (ICL) has demonstrated remarkable success in large language models (LLMs) due to its adaptability and parameter-free nature. However, it also introduces a critical vulnerability to backdoor attacks, where adversaries can manipulate LLM behaviors by simply poisoning a few ICL demonstrations. In this paper, we propose, for the first time, the dual-learning hypothesis, which posits that LLMs simultaneously learn both the task-relevant latent concepts and backdoor latent concepts within poisoned demonstrations, jointly influencing the probability of model outputs. Through theoretical analysis, we derive an upper bound for ICL backdoor effects, revealing that the vulnerability is dominated by the concept preference ratio between the task and the backdoor. Motivated by these findings, we propose ICLShield, a defense mechanism that dynamically adjusts the concept preference ratio. Our method encourages LLMs to select clean demonstrations during the ICL phase by leveraging confidence and similarity scores, effectively mitigating susceptibility to backdoor attacks. Extensive experiments across multiple LLMs and tasks demonstrate that our method achieves state-of-the-art defense effectiveness, significantly outperforming existing approaches (+26. 02% on average). Furthermore, our method exhibits exceptional adaptability and defensive performance even for closed-source models ( e. g. , GPT-4).

AAAI Conference 2025 Conference Paper

Improving Complex Reasoning over Knowledge Graph with Logic-Aware Curriculum Tuning

• Tianle Xia
• Liang Ding
• Guojia Wan
• Yibing Zhan
• Bo Du
• Dacheng Tao

Answering complex queries over incomplete knowledge graphs (KGs) is a challenging job. Most previous works have focused on learning entity/relation embeddings and simulating first-order logic operators with various neural networks. However, they are bottlenecked by the inability to share world knowledge to improve logical reasoning, thus resulting in suboptimal performance. In this paper, we propose a complex reasoning schema over KG upon large language models (LLMs), containing a curriculum-based logical-aware instruction tuning framework, named LACT. Specifically, we augment the arbitrary first-order logical queries via binary tree decomposition, to stimulate the reasoning capability of LLMs. To address the difficulty gap among different types of complex queries, we design a simple and flexible logic-aware curriculum learning framework. Experiments across widely used datasets demonstrate that LACT has substantial improvements~(brings an average +5.5% MRR score) over advanced methods, achieving the new state-of-the-art.

ICML Conference 2025 Conference Paper

Learn from Downstream and Be Yourself in Multimodal Large Language Models Fine-Tuning

• Wenke Huang 0003
• Jian Liang 0003
• Zekun Shi
• Didi Zhu
• Guancheng Wan
• He Li 0054
• Bo Du 0001
• Dacheng Tao

Multimodal Large Language Model (MLLM) has demonstrated strong generalization capabilities across diverse distributions and tasks, largely due to extensive pre-training datasets. Fine-tuning MLLM has become a common practice to improve performance on specific downstream tasks. However, during fine-tuning, MLLM often faces the risk of forgetting knowledge acquired during pre-training, which can result in a decline in generalization abilities. To balance the trade-off between generalization and specialization, we propose measuring the parameter importance for both pre-trained and fine-tuning distributions, based on frozen pre-trained weight magnitude and accumulated fine-tuning gradient values. We further apply an importance-aware weight allocation strategy, selectively updating relatively important parameters for downstream tasks. We conduct empirical evaluations on both image captioning and visual question-answering tasks using various MLLM architectures. The comprehensive experimental analysis demonstrates the effectiveness of the proposed solution, highlighting the efficiency of the crucial modules in enhancing downstream specialization performance while mitigating generalization degradation in MLLM Fine-Tuning.

ICLR Conference 2025 Conference Paper

Learning system dynamics without forgetting

• Xikun Zhang 0002
• Dongjin Song
• Yushan Jiang
• Yixin Chen 0001
• Dacheng Tao

Observation-based trajectory prediction for systems with unknown dynamics is essential in fields such as physics and biology. Most existing approaches are limited to learning within a single system with fixed dynamics patterns. However, many real-world applications require learning across systems with evolving dynamics patterns, a challenge that has been largely overlooked. To address this, we systematically investigate the problem of Continual Dynamics Learning (CDL), examining task configurations and evaluating the applicability of existing techniques, while identifying key challenges. In response, we propose the Mode-switching Graph ODE (MS-GODE) model, which integrates the strengths LG-ODE and sub-network learning with a mode-switching module, enabling efficient learning over varying dynamics. Moreover, we construct a novel benchmark of biological dynamic systems for CDL, Bio-CDL, featuring diverse systems with disparate dynamics and significantly enriching the research field of machine learning for dynamic systems. Our code available at \url{https://github.com/QueuQ/MS-GODE}.

ICML Conference 2025 Conference Paper

LLM Data Selection and Utilization via Dynamic Bi-level Optimization

• Yang Yu 0056
• Kai Han 0002
• Hang Zhou
• Yehui Tang 0001
• Kaiqi Huang
• Yunhe Wang 0001
• Dacheng Tao

While large-scale training data is fundamental for developing capable large language models (LLMs), strategically selecting high-quality data has emerged as a critical approach to enhance training efficiency and reduce computational costs. Current data selection methodologies predominantly rely on static, training-agnostic criteria, failing to account for the dynamic model training and data interactions. In this paper, we propose a new Data Weighting Model (DWM) to adjust the weight of selected data within each batch to achieve a dynamic data utilization during LLM training. Specially, to better capture the dynamic data preference of the trained model, a bi-level optimization framework is implemented to update the weighting model. Our experiments demonstrate that DWM enhances the performance of models trained with randomly-selected data, and the learned weighting model can be transferred to enhance other data selection methods and models of different sizes. Moreover, we further analyze how a model’s data preferences evolve throughout training, providing new insights into the data preference of the model during training.

ICML Conference 2025 Conference Paper

Mastering Massive Multi-Task Reinforcement Learning via Mixture-of-Expert Decision Transformer

• Yilun Kong
• Guozheng Ma
• Qi Zhao
• Haoyu Wang 0018
• Li Shen 0008
• Xueqian Wang 0001
• Dacheng Tao

Despite recent advancements in offline multi-task reinforcement learning (MTRL) have harnessed the powerful capabilities of the Transformer architecture, most approaches focus on a limited number of tasks, with scaling to extremely massive tasks remaining a formidable challenge. In this paper, we first revisit the key impact of task numbers on current MTRL method, and further reveal that naively expanding the parameters proves insufficient to counteract the performance degradation as the number of tasks escalates. Building upon these insights, we propose M3DT, a novel mixture-of-experts (MoE) framework that tackles task scalability by further unlocking the model’s parameter scalability. Specifically, we enhance both the architecture and the optimization of the agent, where we strengthen the Decision Transformer (DT) backbone with MoE to reduce task load on parameter subsets, and introduce a three-stage training mechanism to facilitate efficient training with optimal performance. Experimental results show that, by increasing the number of experts, M3DT not only consistently enhances its performance as model expansion on the fixed task numbers, but also exhibits remarkable task scalability, successfully extending to 160 tasks with superior performance.

NeurIPS Conference 2025 Conference Paper

Merging on the Fly Without Retraining: A Sequential Approach to Scalable Continual Model Merging

• Anke Tang
• Enneng Yang
• Li Shen
• Yong Luo
• Han Hu
• Lefei Zhang
• Bo Du
• Dacheng Tao

Deep model merging represents an emerging research direction that combines multiple fine-tuned models to harness their specialized capabilities across different tasks and domains. Current model merging techniques focus on merging all available models simultaneously, with weight interpolation-based methods being the predominant approach. However, these conventional approaches are not well-suited for scenarios where models become available sequentially, and they often suffer from high memory requirements and potential interference between tasks. In this study, we propose a training-free projection-based continual merging method that processes models sequentially through orthogonal projections of weight matrices and adaptive scaling mechanisms. Our method operates by projecting new parameter updates onto subspaces orthogonal to existing merged parameter updates while using an adaptive scaling mechanism to maintain stable parameter distances, enabling efficient sequential integration of task-specific knowledge. Our approach maintains constant memory complexity to the number of models, minimizes interference between tasks through orthogonal projections, and retains the performance of previously merged models through adaptive task vector scaling. Extensive experiments on CLIP-ViT models demonstrate that our method achieves a 5-8% average accuracy improvement while maintaining robust performance in different task orderings. Code is publicly available at https: //github. com/tanganke/opcm.

AAAI Conference 2025 Conference Paper

Modeling All Response Surfaces in One for Conditional Search Spaces

• Jiaxing Li
• Wei Liu
• Chao Xue
• Yibing Zhan
• Xiaoxing Wang
• Weifeng Liu
• Dacheng Tao

Bayesian Optimization (BO) is a sample-efficient black-box optimizer commonly used in search spaces where hyperparameters are independent. However, in many practical AutoML scenarios, there will be dependencies among hyperparameters, forming a conditional search space, which can be partitioned into structurally distinct subspaces. The structure and dimensionality of hyperparameter configurations vary across these subspaces, challenging the application of BO. Some previous BO works have proposed solutions to develop multiple Gaussian Process models in these subspaces. However, these approaches tend to be inefficient as they require a substantial number of observations to guarantee each GP's performance and cannot capture relationships between hyperparameters across different subspaces. To address these issues, this paper proposes a novel approach to model the response surfaces of all subspaces in one, which can model the relationships between hyperparameters elegantly via a self-attention mechanism. Concretely, we design a structure-aware hyperparameter embedding to preserve the structural information. Then, we introduce an attention-based deep feature extractor, capable of projecting configurations with different structures from various subspaces into a unified feature space, where the response surfaces can be formulated using a single standard Gaussian Process. The empirical results on a simulation function, various real-world tasks, and HPO-B benchmark demonstrate that our proposed approach improves the efficacy and efficiency of BO within conditional search spaces.

NeurIPS Conference 2025 Conference Paper

Mulberry: Empowering MLLM with o1-like Reasoning and Reflection via Collective Monte Carlo Tree Search

• Huanjin Yao
• Jiaxing Huang
• Wenhao Wu
• Jingyi Zhang
• Yibo Wang
• Shunyu Liu
• Yingjie Wang
• YuXin Song

In this work, we aim to develop an MLLM that understands and solves questions by learning to create each intermediate step of the reasoning involved till the final answer. To this end, we propose Collective Monte Carlo Tree Search (CoMCTS), a new learning-to-reason method for MLLMs, which introduces the concept of collective learning into ``tree search'' for effective and efficient reasoning-path searching and learning. The core idea of CoMCTS is to leverage collective knowledge from multiple models to collaboratively conjecture, search and identify effective reasoning paths toward correct answers via four iterative operations including Expansion, Simulation and Error Positioning, Backpropagation, and Selection. Using CoMCTS, we construct Mulberry-260k, a multimodal dataset with a tree of rich, explicit and well-defined reasoning nodes for each question. With Mulberry-260k, we perform collective SFT to train our model, Mulberry, a series of MLLMs with o1-like step-by-step Reasoning and Reflection capabilities. Extensive experiments demonstrate the superiority of our proposed methods on various benchmarks. Code is available at https: //github. com/HJYao00/Mulberry.

ICML Conference 2025 Conference Paper

Multinoulli Extension: A Lossless Yet Effective Probabilistic Framework for Subset Selection over Partition Constraints

• Qixin Zhang 0001
• Wei Huang
• Can Jin
• Puning Zhao
• Yao Shu
• Li Shen 0008
• Dacheng Tao

Identifying the most representative subset for a close-to-submodular objective while satisfying the predefined partition constraint is a fundamental task with numerous applications in machine learning. However, the existing distorted local-search methods are often hindered by their prohibitive query complexities and the rigid requirement for prior knowledge of difficult-to-obtain structural parameters. To overcome these limitations, we introduce a novel algorithm titled Multinoulli-SCG, which not only is parameter-free, but also can achieve the same approximation guarantees as the distorted local-search methods with significantly fewer function evaluations. The core of our Multinoulli-SCG algorithm is an innovative continuous-relaxation framework named Multinoulli Extension( ME ), which can effectively convert the discrete subset selection problem subject to partition constraints into a solvable continuous maximization focused on learning the optimal multinoulli priors across the considered partition. In sharp contrast with the well-established multi-linear extension for submodular subset selection, a notable advantage of our proposed ME is its intrinsic capacity to provide a lossless rounding scheme for any set function. Finally, we validate the practical efficacy of our proposed algorithms by applying them to video summarization, bayesian A-optimal design and coverage maximization.

ICLR Conference 2025 Conference Paper

Near-Optimal Online Learning for Multi-Agent Submodular Coordination: Tight Approximation and Communication Efficiency

• Qixin Zhang 0001
• Zongqi Wan
• Yu Yang 0001
• Li Shen 0008
• Dacheng Tao

Coordinating multiple agents to collaboratively maximize submodular functions in unpredictable environments is a critical task with numerous applications in machine learning, robot planning and control. The existing approaches, such as the OSG algorithm, are often hindered by their poor approximation guarantees and the rigid requirement for a fully connected communication graph. To address these challenges, we firstly present a $\textbf{MA-OSMA}$ algorithm, which employs the multi-linear extension to transfer the discrete submodular maximization problem into a continuous optimization, thereby allowing us to reduce the strict dependence on a complete graph through consensus techniques. Moreover, $\textbf{MA-OSMA}$ leverages a novel surrogate gradient to avoid sub-optimal stationary points. To eliminate the computationally intensive projection operations in $\textbf{MA-OSMA}$, we also introduce a projection-free $\textbf{MA-OSEA}$ algorithm, which effectively utilizes the KL divergence by mixing a uniform distribution. Theoretically, we confirm that both algorithms achieve a regret bound of $\widetilde{O}(\sqrt{\frac{C_{T}T}{1-\beta}})$ against a $(\frac{1-e^{-c}}{c})$-approximation to the best comparator in hindsight, where $C_{T}$ is the deviation of maximizer sequence, $\beta$ is the spectral gap of the network and $c$ is the joint curvature of submodular objectives. This result significantly improves the $(\frac{1}{1+c})$-approximation provided by the state-of-the-art OSG algorithm. Finally, we demonstrate the effectiveness of our proposed algorithms through simulation-based multi-target tracking.

ICML Conference 2025 Conference Paper

Network Sparsity Unlocks the Scaling Potential of Deep Reinforcement Learning

• Guozheng Ma
• Lu Li
• Zilin Wang 0002
• Li Shen 0008
• Pierre-Luc Bacon
• Dacheng Tao

Effectively scaling up deep reinforcement learning models has proven notoriously difficult due to network pathologies during training, motivating various targeted interventions such as periodic reset and architectural advances such as layer normalization. Instead of pursuing more complex modifications, we show that introducing static network sparsity alone can unlock further scaling potential beyond their dense counterparts with state-of-the-art architectures. This is achieved through simple one-shot random pruning, where a predetermined percentage of network weights are randomly removed once before training. Our analysis reveals that, in contrast to naively scaling up dense DRL networks, such sparse networks achieve both higher parameter efficiency for network expressivity and stronger resistance to optimization challenges like plasticity loss and gradient interference. We further extend our evaluation to visual and streaming RL scenarios, demonstrating the consistent benefits of network sparsity.

ICLR Conference 2025 Conference Paper

NoVo: Norm Voting off Hallucinations with Attention Heads in Large Language Models

• Zheng Yi Ho
• Siyuan Liang 0004
• Sen Zhang 0006
• Yibing Zhan
• Dacheng Tao

Hallucinations in Large Language Models (LLMs) remain a major obstacle, particularly in high-stakes applications where factual accuracy is critical. While representation editing and reading methods have made strides in reducing hallucinations, their heavy reliance on specialised tools and training on in-domain samples, makes them difficult to scale and prone to overfitting. This limits their accuracy gains and generalizability to diverse datasets. This paper presents a lightweight method, Norm Voting (NoVo), which harnesses the untapped potential of attention head norms to dramatically enhance factual accuracy in zero-shot multiple-choice questions (MCQs). NoVo begins by automatically selecting truth-correlated head norms with an efficient, inference-only algorithm using only 30 random samples, allowing NoVo to effortlessly scale to diverse datasets. Afterwards, selected head norms are employed in a simple voting algorithm, which yields significant gains in prediction accuracy. On TruthfulQA MC1, NoVo surpasses the current state-of-the-art and all previous methods by an astounding margin---at least 19 accuracy points. NoVo demonstrates exceptional generalization to 20 diverse datasets, with significant gains in over 90\% of them, far exceeding all current representation editing and reading methods. NoVo also reveals promising gains to finetuning strategies and building textual adversarial defence. NoVo's effectiveness with head norms opens new frontiers in LLM interpretability, robustness and reliability. Our code is available at: https://github.com/hozhengyi/novo

NeurIPS Conference 2025 Conference Paper

On Geometry-Enhanced Parameter-Efficient Fine-Tuning for 3D Scene Segmentation

• Liyao Tang
• Zhe Chen
• Dacheng Tao

The emergence of large-scale pre-trained point cloud models has significantly advanced 3D scene understanding, but adapting these models to specific downstream tasks typically demands full fine-tuning, incurring high computational and storage costs. Parameter-efficient fine-tuning (PEFT) techniques, successful in natural language processing and 2D vision tasks, would underperform when naively applied to 3D point cloud models due to significant geometric and spatial distribution shifts. Existing PEFT methods commonly treat points as orderless tokens, neglecting important local spatial structures and global geometric contexts in 3D modeling. To bridge this gap, we introduce the Geometric Encoding Mixer (GEM), a novel geometry-aware PEFT module specifically designed for 3D point cloud transformers. GEM explicitly integrates fine-grained local positional encodings with a lightweight latent attention mechanism to capture comprehensive global context, thereby effectively addressing the spatial and geometric distribution mismatch. Extensive experiments demonstrate that GEM achieves performance comparable to or sometimes even exceeding full fine-tuning, while only updating 1. 6% of the model's parameters, fewer than other PEFT methods. With significantly reduced training time and memory requirements, our approach thus sets a new benchmark for efficient, scalable, and geometry-aware fine-tuning of large-scale 3D point cloud models. Code is available at https: //github. com/LiyaoTang/GEM.

ICLR Conference 2025 Conference Paper

Open-Vocabulary Customization from CLIP via Data-Free Knowledge Distillation

• Yongxian Wei
• Zixuan Hu
• Li Shen 0008
• Zhenyi Wang 0001
• Chun Yuan 0003
• Dacheng Tao

Vision-language models such as CLIP have demonstrated strong zero-shot performance, but their considerable size and inefficient inference limit customizable deployment for users. While knowledge distillation is a solution, it still requires the original data, which is not always available due to copyrights and privacy concerns. For many users seeking open-vocabulary customization, Data-Free Knowledge Distillation (DFKD) emerges as a promising direction. Upon rethinking DFKD, we find that existing methods fail on CLIP due to their heavy reliance on BatchNorm layers, which are unexpectedly unusable in CLIP. Based on our findings, we adopt image-text matching to achieve DFKD for CLIP, enabling customization based on arbitrary class texts. This involves (i) inversing a surrogate dataset from CLIP based on text prompts; and (ii) distilling a student model from CLIP using the surrogate dataset. Specifically, we introduce style dictionary diversification to enhance the diversity of synthetic images. To prevent uncontrollable semantics introduced by diversification, we propose a class consistency maintaining strategy to ensure the consistency of synthetic images. Based on synthetic images with various styles, we further propose meta knowledge distillation to train the student model with good generalization ability. Moreover, we introduce a simple yet effective method to enable customization based on few example images. Comprehensive experiments showcase the superiority of our approach across twelve customized tasks, achieving a 9.33\% improvement compared to existing DFKD methods.

NeurIPS Conference 2025 Conference Paper

Panacea: Mitigating Harmful Fine-tuning for Large Language Models via Post-fine-tuning Perturbation

• Yibo Wang
• Tiansheng Huang
• Li Shen
• Huanjin Yao
• Haotian Luo
• Rui Liu
• Naiqiang Tan
• Jiaxing Huang

Harmful fine-tuning attack introduces significant security risks to the fine-tuning services. Main-stream defenses aim to vaccinate the model such that the later harmful fine-tuning attack is less effective. However, our evaluation results show that such defenses are fragile-- with a few fine-tuning steps, the model still can learn the harmful knowledge. To this end, we do further experiment and find that an embarrassingly simple solution-- adding purely random perturbations to the fine-tuned model, can recover the model from harmful behaviors, though it leads to a degradation in the model’s fine-tuning performance. To address the degradation of fine-tuning performance, we further propose \methodname, which optimizes an adaptive perturbation that will be applied to the model after fine-tuning. \methodname maintains model's safety alignment performance without compromising downstream fine-tuning performance. Comprehensive experiments are conducted on different harmful ratios, fine-tuning tasks and mainstream LLMs, where the average harmful scores are reduced by up-to 21. 2%, while maintaining fine-tuning performance. As a by-product, we analyze the adaptive perturbation and show that different layers in various LLMs have distinct safety coefficients. Source code available at https: //github. com/w-yibo/Panacea.

AAMAS Conference 2025 Conference Paper

Prompt Tuning with Diffusion for Few-Shot Pre-trained Policy Generalization

• Shengchao Hu
• Wanru Zhao
• Weixiong Lin
• Li Shen
• Ya Zhang
• Dacheng Tao

Offline reinforcement learning (RL) methods harness previous experiences to derive an optimal policy, forming the foundation for pretrained large-scale models (PLMs). When adapting to novel tasks, PLMs leverage expert trajectories as prompts to accelerate adaptation. While various prompt-tuning techniques aim to improve prompt quality, their effectiveness is often limited by initialization constraints, restricting exploration and potentially leading to suboptimal solutions. To eliminate dependence on the initial prompt, we reframe prompt-tuning as conditional generative modeling, where prompts are generated from random noise. Our proposed Prompt Diffuser employs a conditional diffusion model to generate high-quality prompts. Central to our framework is trajectory reconstruction and the seamless integration of downstream task guidance during training. Experimental results validate Prompt Diffuser’s effectiveness, demonstrating strong performance in meta-RL tasks.

TMLR Journal 2025 Journal Article

QPO: Query-dependent Prompt Optimization via Multi-Loop Offline Reinforcement Learning

• Yilun Kong
• Hangyu Mao
• Zhao Qi
• Bin Zhang
• Jingqing Ruan
• Li Shen
• Yongzhe Chang
• Xueqian Wang

Prompt engineering has demonstrated remarkable success in enhancing the performance of large language models (LLMs) across diverse tasks. However, most existing prompt optimization methods only focus on the task-level performance, overlooking the importance of query-preferred prompts, which leads to suboptimal performances. Additionally, these methods rely heavily on frequent interactions with LLMs to obtain feedback for guiding the optimization process, incurring substantial redundant interaction costs. In this paper, we introduce Query-dependent Prompt Optimization ($\textbf{QPO}$), which leverages multi-loop offline reinforcement learning to iteratively fine-tune a small pretrained language model to generate optimal prompts tailored to the input queries, thus significantly improving the prompting effect on the large target LLM. We derive insights from offline prompting demonstration data, which already exists in large quantities as a by-product of benchmarking diverse prompts on open-sourced tasks, thereby circumventing the expenses of online interactions. Furthermore, we continuously augment the offline dataset with the generated prompts in each loop, as the prompts from the fine-tuned model are supposed to outperform the source prompts in the original dataset. These iterative loops bootstrap the model towards generating optimal prompts. Experiments on various LLM scales and diverse NLP and math tasks demonstrate the efficacy and cost-efficiency of our method in both zero-shot and few-shot scenarios.

NeurIPS Conference 2025 Conference Paper

R1-ShareVL: Incentivizing Reasoning Capabilities of Multimodal Large Language Models via Share-GRPO

• Huanjin Yao
• Qixiang Yin
• Jingyi Zhang
• Min Yang
• Yibo Wang
• Wenhao Wu
• Fei Su
• Li Shen

In this work, we aim to incentivize the reasoning ability of Multimodal Large Language Models (MLLMs) via reinforcement learning (RL) and develop an effective approach that mitigates the sparse reward and advantage vanishing issues during RL. To this end, we propose Share-GRPO, a novel RL approach that tackle these issues by exploring and sharing diverse reasoning trajectories over expanded question space. Specifically, Share-GRPO first expands the question space for a given question via data transformation techniques, and then encourages MLLM to effectively explore diverse reasoning trajectories over the expanded question space and shares the discovered reasoning trajectories across the expanded questions during RL. In addition, Share-GRPO also shares reward information during advantage computation, which estimates solution advantages hierarchically across and within question variants, allowing more accurate estimation of relative advantages and improving the stability of policy training. Extensive evaluations over 6 widely-used reasoning benchmarks showcase the superior performance of our method. Code is available at https: //github. com/HJYao00/R1-ShareVL.

ICML Conference 2025 Conference Paper

Retrieval-Augmented Perception: High-resolution Image Perception Meets Visual RAG

• Wenbin Wang
• Yongcheng Jing
• Liang Ding 0006
• Yingjie Wang
• Li Shen 0008
• Yong Luo 0002
• Bo Du 0001
• Dacheng Tao

High-resolution (HR) image perception remains a key challenge in multimodal large language models (MLLMs). To drive progress beyond the limits of heuristic methods, this paper advances HR perception capabilities of MLLMs by harnessing cutting-edge long-context techniques such as retrieval-augmented generation (RAG). Towards this end, this paper presents the first study exploring the use of RAG to address HR perception challenges. Specifically, we propose Retrieval-Augmented Perception (RAP), a training-free framework that retrieves and fuses relevant image crops while preserving spatial context using the proposed Spatial-Awareness Layout. To accommodate different tasks, the proposed Retrieved-Exploration Search (RE-Search) dynamically selects the optimal number of crops based on model confidence and retrieval scores. Experimental results on HR benchmarks demonstrate the significant effectiveness of RAP, with LLaVA-v1. 5-13B achieving a 43% improvement on $V^*$ Bench and 19% on HR-Bench. Code is available at https: //github. com/DreamMr/RAP.

ICML Conference 2025 Conference Paper

Safety Reasoning with Guidelines

• Haoyu Wang 0018
• Zeyu Qin
• Li Shen 0008
• Xueqian Wang 0001
• Dacheng Tao
• Minhao Cheng

Training safe LLMs remains a critical challenge. The most widely used method, Refusal Training (RT), struggles to generalize against various Out-of-Distribution (OOD) jailbreaking attacks. Although various advanced methods have been proposed to address this issue, we instead question whether OOD attacks inherently surpass the capability of vanilla RT. Evaluations using Best-of-N (BoN) reveal significant safety improvements as N increases, indicating models possess adequate latent safety knowledge but RT fails to consistently elicit it under OOD scenarios. Further domain adaptation analysis reveals that direct RT causes reliance on superficial shortcuts, resulting in non-generalizable representation mappings. Inspired by our findings, we propose training model to perform safety reasoning for each query. Specifically, we synthesize reasoning supervision aligned with specified guidelines that reflect diverse perspectives on safety knowledge. This encourages model to engage in deeper reasoning, explicitly eliciting and utilizing latent safety knowledge for each query. Extensive experiments show that our method significantly improves model generalization against OOD attacks.

NeurIPS Conference 2025 Conference Paper

SEGA: Shaping Semantic Geometry for Robust Hashing under Noisy Supervision

• Yiyang Gu
• Bohan Wu
• Qinghua Ran
• Rong-Cheng Tu
• Xiao Luo
• Zhiping Xiao
• Wei Ju
• Dacheng Tao

This paper studies the problem of learning hash codes from noisy supervision, which is a practical yet challenging task. This problem is important in extensive real-world applications such as image retrieval and cross-modal retrieval. However, most of the existing methods focus on label denoising to address this problem, but ignore the geometric structure of the hash space, which is critical for learning stable hash codes. Towards this end, this paper proposes a novel framework named Semantic Geometry Shaping (SEGA) that explicitly refines the semantic geometry of hash space. Specifically, we first learn dynamic class prototypes as semantic anchors and cluster hash embeddings around these prototypes to keep structural stability. We then leverage both the energy of predicted distributions and structure-based divergence to estimate the uncertainty of instances and calibrate the supervision in a soft manner. Moreover, we introduce structure-aware interpolation to improve the class boundaries. To verify the effectiveness of our design, we give the theoretical analysis for the proposed framework. Experiments on a range of widely-used retrieval datasets justify the superiority of our SEGA over extensive strong baselines under noisy supervision.

NeurIPS Conference 2025 Conference Paper

Self-Verification Provably Prevents Model Collapse in Recursive Synthetic Training

• Shi Fu
• Yingjie Wang
• Yuzhu Chen
• Li Shen
• Dacheng Tao

Large generative models are increasingly trained on synthetic data from earlier generations, raising concerns about model collapse, a progressive performance decline consistently observed in empirical studies. However, theoretical understanding of recursive training dynamics and their failure modes remains limited. In this work, we theoretically show that recursive training inherently leads to exponential error growth unless mitigated by sufficient real data. Addressing the growing scarcity of real data, we introduce a self-verification mechanism enabling models to filter their outputs based on internal confidence scores without external validation. Through rigorous analysis, we derive finite-sample error bounds demonstrating that self-verification alone can prevent collapse, even in fully synthetic training regimes. Our theoretical framework extends to large language models (LLMs), characterizing the conditions under which recursive training can maintain stability without performance degradation.

NeurIPS Conference 2025 Conference Paper

SeRL: Self-play Reinforcement Learning for Large Language Models with Limited Data

• Wenkai Fang
• Shunyu Liu
• Yang Zhou
• Kongcheng Zhang
• Tongya Zheng
• Kaixuan Chen
• Mingli Song
• Dacheng Tao

Recent advances have demonstrated the effectiveness of Reinforcement Learning (RL) in improving the reasoning capabilities of Large Language Models (LLMs). However, existing works inevitably rely on high-quality instructions and verifiable rewards for effective training, both of which are often difficult to obtain in specialized domains. In this paper, we propose Self-play Reinforcement Learning (SeRL) to bootstrap LLM training with limited initial data. Specifically, SeRL comprises two complementary modules: self-instruction and self-rewarding. The former module generates additional instructions based on the available data at each training step, employing comprehensive online filtering strategies to ensure instruction quality, diversity, and difficulty. The latter module introduces a simple yet effective majority-voting mechanism to estimate response rewards for additional instructions, eliminating the need for external annotations. Finally, SeRL performs conventional RL based on the generated data, facilitating iterative self-play learning. Extensive experiments on various reasoning benchmarks and across different LLM backbones demonstrate that the proposed SeRL yields results superior to its counterparts and achieves performance on par with those obtained by high-quality data with verifiable rewards. Our code is available at https: //github. com/wantbook-book/SeRL.

NeurIPS Conference 2025 Conference Paper

SPAZER: Spatial-Semantic Progressive Reasoning Agent for Zero-shot 3D Visual Grounding

• Zhao Jin
• Rong-Cheng Tu
• Jingyi Liao
• Wenhao Sun
• Xiao Luo
• Shunyu Liu
• Dacheng Tao

3D Visual Grounding (3DVG) aims to localize target objects within a 3D scene based on natural language queries. To alleviate the reliance on costly 3D training data, recent studies have explored zero-shot 3DVG by leveraging the extensive knowledge and powerful reasoning capabilities of pre-trained LLMs and VLMs. However, existing paradigms tend to emphasize either spatial (3D-based) or semantic (2D-based) understanding, limiting their effectiveness in complex real-world applications. In this work, we introduce SPAZER — a VLM-driven agent that combines both modalities in a progressive reasoning framework. It first holistically analyzes the scene and produces a 3D rendering from the optimal viewpoint. Based on this, anchor-guided candidate screening is conducted to perform a coarse-level localization of potential objects. Furthermore, leveraging retrieved relevant 2D camera images, 3D-2D joint decision-making is efficiently performed to determine the best-matching object. By bridging spatial and semantic reasoning neural streams, SPAZER achieves robust zero-shot grounding without training on 3D-labeled data. Extensive experiments on ScanRefer and Nr3D benchmarks demonstrate that SPAZER significantly outperforms previous state-of-the-art zero-shot methods, achieving notable gains of $\mathbf{9. 0\}$% and $\mathbf{10. 9\}$% in accuracy.

NeurIPS Conference 2025 Conference Paper

T2V-OptJail: Discrete Prompt Optimization for Text-to-Video Jailbreak Attacks

• Jiayang Liu
• Siyuan Liang
• Shiqian Zhao
• Rong-Cheng Tu
• Wenbo Zhou
• Aishan Liu
• Dacheng Tao
• Siew Kei Lam

In recent years, fueled by the rapid advancement of diffusion models, text-to-video (T2V) generation models have achieved remarkable progress, with notable examples including Pika, Luma, Kling, and Open-Sora. Although these models exhibit impressive generative capabilities, they also expose significant security risks due to their vulnerability to jailbreak attacks, where the models are manipulated to produce unsafe content such as pornography, violence, or discrimination. Existing works such as T2VSafetyBench provide preliminary benchmarks for safety evaluation, but lack systematic methods for thoroughly exploring model vulnerabilities. To address this gap, we are the first to formalize the T2V jailbreak attack as a discrete optimization problem and propose a joint objective-based optimization framework, called \emph{T2V-OptJail}. This framework consists of two key optimization goals: bypassing the built-in safety filtering mechanisms to increase the attack success rate, preserving semantic consistency between the adversarial prompt and the unsafe input prompt, as well as between the generated video and the unsafe input prompt, to enhance content controllability. In addition, we introduce an iterative optimization strategy guided by prompt variants, where multiple semantically equivalent candidates are generated in each round, and their scores are aggregated to robustly guide the search toward optimal adversarial prompts. We conduct large-scale experiments on several T2V models, covering both open-source models (\textit{e. g. }, Open-Sora) and real commercial closed-source models (\textit{e. g. }, Pika, Luma, Kling). The experimental results show that the proposed method improves 11. 4\% and 10. 0\% over the existing state-of-the-art method (SoTA) in terms of attack success rate assessed by GPT-4, attack success rate assessed by human accessors, respectively, verifying the significant advantages of the method in terms of attack effectiveness and content control. This study reveals the potential abuse risk of the semantic alignment mechanism in the current T2V model and provides a basis for the design of subsequent jailbreak defense methods.

NeurIPS Conference 2025 Conference Paper

Tackling Continual Offline RL through Selective Weights Activation on Aligned Spaces

• Jifeng Hu
• Sili Huang
• Li Shen
• Zhejian Yang
• Shengchao Hu
• Shisong Tang
• Hechang Chen
• Lichao Sun

Continual offline reinforcement learning (CORL) has shown impressive ability in diffusion-based continual learning systems by modeling the joint distributions of trajectories. However, most research only focuses on limited continual task settings where the tasks have the same observation and action space, which deviates from the realistic demands of training agents in various environments. In view of this, we propose Vector-Quantized Continual Diffuser, named VQ-CD, to break the barrier of different spaces between various tasks. Specifically, our method contains two complementary sections, where the quantization spaces alignment provides a unified basis for the selective weights activation. In the quantized spaces alignment, we leverage vector quantization to align the different state and action spaces of various tasks, facilitating continual training in the same space. Then, we propose to leverage a unified diffusion model attached by the inverse dynamic model to master all tasks by selectively activating different weights according to the task-related sparse masks. Finally, we conduct extensive experiments on 15 continual learning (CL) tasks, including conventional CL task settings (identical state and action spaces) and general CL task settings (various state and action spaces). Compared with 17 baselines, our method reaches the SOTA performance.

ICML Conference 2025 Conference Paper

The Energy Loss Phenomenon in RLHF: A New Perspective on Mitigating Reward Hacking

• Yuchun Miao
• Sen Zhang 0006
• Liang Ding 0006
• Yuqi Zhang 0002
• Lefei Zhang
• Dacheng Tao

This work identifies the Energy Loss Phenomenon in Reinforcement Learning from Human Feedback (RLHF) and its connection to reward hacking. Specifically, energy loss in the final layer of a Large Language Model (LLM) gradually increases during the RL process, with an excessive increase in energy loss characterizing reward hacking. Beyond empirical analysis, we further provide a theoretical foundation by proving that, under mild conditions, the increased energy loss reduces the upper bound of contextual relevance in LLMs, which is a critical aspect of reward hacking as the reduced contextual relevance typically indicates overfitting to reward model-favored patterns in RL. To address this issue, we propose an Energy loss-aware PPO algorithm (EPPO) which penalizes the increase in energy loss in the LLM’s final layer during reward calculation to prevent excessive energy loss, thereby mitigating reward hacking. We theoretically show that EPPO can be conceptually interpreted as an entropy-regularized RL algorithm, which provides deeper insights into its effectiveness. Extensive experiments across various LLMs and tasks demonstrate the commonality of the energy loss phenomenon, as well as the effectiveness of EPPO in mitigating reward hacking and improving RLHF performance.

NeurIPS Conference 2025 Conference Paper

UltraVideo: High-Quality UHD Video Dataset with Comprehensive Captions

• Xue zhucun
• Jiangning Zhang
• Teng Hu
• Haoyang He
• Yinan Chen
• Yuxuan Cai
• Yabiao Wang
• Chengjie Wang

The quality of the video dataset (image quality, resolution, and fine-grained caption) greatly influences the performance of the video generation model. %The growing demand for video applications sets higher requirements for high-quality video generation models. %For example, the generation of movie-level Ultra-High Definition (UHD) videos and the creation of 4K short video content. %However, the existing public datasets cannot support related research and applications. %In this paper, we first propose a high-quality open-sourced UHD-4K (22. 4\% of which are 8K) text-to-video dataset named UltraVideo, which contains a wide range of topics (more than 100 kinds), and each video has 9 structured captions with one summarized caption (average of 824 words). %Specifically, we carefully design a highly automated curation process with four stages to obtain the final high-quality dataset: i) collection of diverse and high-quality video clips. ii statistical data filtering. iii) model-based data purification. iv) generation of comprehensive, structured captions. %In addition, we expand Wan to UltraWan-1K/-4K, which can natively generate high-quality 1K/4K videos with more consistent text controllability, demonstrating the effectiveness of our data curation. %We believe that this work can make a significant contribution to future research on UHD video generation. UltraVideo dataset and UltraWan models are available at https: //xzc-zju. github. io/projects/UltraVideo.

NeurIPS Conference 2025 Conference Paper

Value-Guided Decision Transformer: A Unified Reinforcement Learning Framework for Online and Offline Settings

• Hongling Zheng
• Li Shen
• Yong Luo
• Deheng Ye
• Shuhan Xu
• Bo Du
• Jialie Shen
• Dacheng Tao

The Conditional Sequence Modeling (CSM) paradigm, benefiting from the transformer's powerful distribution modeling capabilities, has demonstrated considerable promise in Reinforcement Learning (RL) tasks. However, much of the work has focused on applying CSM to single online or offline settings, with the general architecture rarely explored. Additionally, existing methods primarily focus on deterministic trajectory modeling, overlooking the randomness of state transitions and the diversity of future trajectory distributions. Fortunately, value-based methods offer a viable solution for CSM, further bridging the potential gap between offline and online RL. In this paper, we propose Value-Guided Decision Transformer (VDT), which leverages value functions to perform advantage-weighting and behavior regularization on the Decision Transformer (DT), guiding the policy toward upper-bound optimal decisions during the offline training phase. In the online tuning phase, VDT further integrates value-based policy improvement with behavior cloning under the CSM architecture through limited interaction and data collection, achieving performance improvement within minimal timesteps. The predictive capability of value functions for future returns is also incorporated into the sampling process. Our method achieves competitive performance on various standard RL benchmarks, providing a feasible solution for developing CSM architectures in general scenarios. Code is available at here.

NeurIPS Conference 2025 Conference Paper

VORTA: Efficient Video Diffusion via Routing Sparse Attention

• Wenhao Sun
• Rong-Cheng Tu
• Yifu Ding
• Jingyi Liao
• Zhao Jin
• Shunyu Liu
• Dacheng Tao

Video diffusion transformers have achieved remarkable progress in high-quality video generation, but remain computationally expensive due to the quadratic complexity of attention over high-dimensional video sequences. Recent acceleration methods enhance the efficiency by exploiting the local sparsity of attention scores; yet they often struggle with accelerating the long-range computation. To address this problem, we propose VORTA, an acceleration framework with two novel components: 1) a sparse attention mechanism that efficiently captures long-range dependencies, and 2) a routing strategy that adaptively replaces full 3D attention with specialized sparse attention variants. VORTA achieves an end-to-end speedup $1. 76\times$ without loss of quality on VBench. Furthermore, it can seamlessly integrate with various other acceleration methods, such as model caching and step distillation, reaching up to speedup $14. 41\times$ with negligible performance degradation. VORTA demonstrates its efficiency and enhances the practicality of video diffusion transformers in real-world settings. Codes and weights are available at https: //github. com/wenhao728/VORTA.

NeurIPS Conference 2024 Conference Paper

A-FedPD: Aligning Dual-Drift is All Federated Primal-Dual Learning Needs

• Yan Sun
• Li Shen
• Dacheng Tao

As a popular paradigm for juggling data privacy and collaborative training, federated learning (FL) is flourishing to distributively process the large scale of heterogeneous datasets on edged clients. Due to bandwidth limitations and security considerations, it ingeniously splits the original problem into multiple subproblems to be solved in parallel, which empowers primal dual solutions to great application values in FL. In this paper, we review the recent development of classical federated primal dual methods and point out a serious common defect of such methods in non-convex scenarios, which we say is a ``dual drift'' caused by dual hysteresis of those longstanding inactive clients under partial participation training. To further address this problem, we propose a novel Aligned Federated Primal Dual (A-FedPD) method, which constructs virtual dual updates to align global consensus and local dual variables for those protracted unparticipated local clients. Meanwhile, we provide a comprehensive analysis of the optimization and generalization efficiency for the A-FedPD method on smooth non-convex objectives, which confirms its high efficiency and practicality. Extensive experiments are conducted on several classical FL setups to validate the effectiveness of our proposed method.

ICLR Conference 2024 Conference Paper

AdaMerging: Adaptive Model Merging for Multi-Task Learning

• Enneng Yang
• Zhenyi Wang 0001
• Li Shen 0008
• Shiwei Liu 0003
• Guibing Guo
• Xingwei Wang 0001
• Dacheng Tao

Multi-task learning (MTL) aims to empower a model to tackle multiple tasks simultaneously. A recent development known as task arithmetic has revealed that several models, each fine-tuned for distinct tasks, can be directly merged into a single model to execute MTL without necessitating a retraining process using the initial training data. Nevertheless, this direct addition of models often leads to a significant deterioration in the overall performance of the merged model. This decline occurs due to potential conflicts and intricate correlations among the multiple tasks. Consequently, the challenge emerges of how to merge pre-trained models more effectively without using their original training data. This paper introduces an innovative technique called Adaptive Model Merging (AdaMerging). This approach aims to autonomously learn the coefficients for model merging, either in a task-wise or layer-wise manner, without relying on the original training data. Specifically, our AdaMerging method operates as an automatic, unsupervised task arithmetic scheme. It leverages entropy minimization on unlabeled test samples from the multi-task setup as a surrogate objective function to iteratively refine the merging coefficients of the multiple models. Our experimental findings across eight tasks demonstrate the efficacy of the AdaMerging scheme we put forth. Compared to the current state-of-the-art (SOTA) task arithmetic merging scheme, AdaMerging showcases a remarkable 11\% improvement in performance. Notably, AdaMerging also exhibits superior generalization capabilities when applied to unseen downstream tasks. Furthermore, it displays a significantly enhanced robustness to data distribution shifts that may occur during the testing phase.

UAI Conference 2024 Conference Paper

Adaptive Time-Stepping Schedules for Diffusion Models

• Yuzhu Chen
• Fengxiang He
• Shi Fu
• Xinmei Tian 0001
• Dacheng Tao

This paper studies how to tune the stepping schedule in diffusion models, which is mostly fixed in current practice, lacking theoretical foundations and assurance of optimal performance at the chosen discretization points. In this paper, we advocate the use of adaptive time-stepping schedules and design two algorithms with an optimized sampling error bound $EB$: (1) for continuous diffusion, we treat $EB$ as the loss function to discretization points and run gradient descent to adjust them; and (2) for discrete diffusion, we propose a greedy algorithm that adjusts only one discretization point to its best position in each iteration. We conducted extensive experiments that show (1) improved generation ability in well-trained models, and (2) premature though usable generation ability in under-trained models. The code is submitted and will be released publicly.

ICML Conference 2024 Conference Paper

Confronting Reward Overoptimization for Diffusion Models: A Perspective of Inductive and Primacy Biases

• Ziyi Zhang 0001
• Sen Zhang 0006
• Yibing Zhan
• Yong Luo 0002
• Yonggang Wen 0001
• Dacheng Tao

Bridging the gap between diffusion models and human preferences is crucial for their integration into practical generative workflows. While optimizing downstream reward models has emerged as a promising alignment strategy, concerns arise regarding the risk of excessive optimization with learned reward models, which potentially compromises ground-truth performance. In this work, we confront the reward overoptimization problem in diffusion model alignment through the lenses of both inductive and primacy biases. We first identify a mismatch between current methods and the temporal inductive bias inherent in the multi-step denoising process of diffusion models, as a potential source of reward overoptimization. Then, we surprisingly discover that dormant neurons in our critic model act as a regularization against reward overoptimization while active neurons reflect primacy bias. Motivated by these observations, we propose Temporal Diffusion Policy Optimization with critic active neuron Reset (TDPO-R), a policy gradient algorithm that exploits the temporal inductive bias of diffusion models and mitigates the primacy bias stemming from active neurons. Empirical results demonstrate the superior efficacy of our methods in mitigating reward overoptimization. Code is avaliable at https: //github. com/ZiyiZhang27/tdpo.

ICLR Conference 2024 Conference Paper

Convergence of Bayesian Bilevel Optimization

• Shi Fu
• Fengxiang He
• Xinmei Tian 0001
• Dacheng Tao

This paper presents the first theoretical guarantee for Bayesian bilevel optimization (BBO) that we term for the prevalent bilevel framework combining Bayesian optimization at the outer level to tune hyperparameters, and the inner-level stochastic gradient descent (SGD) for training the model. We prove sublinear regret bounds suggesting simultaneous convergence of the inner-level model parameters and outer-level hyperparameters to optimal configurations for generalization capability. A pivotal, technical novelty in the proofs is modeling the excess risk of the SGD-trained parameters as evaluation noise during Bayesian optimization. Our theory implies the inner unit horizon, defined as the number of SGD iterations, shapes the convergence behavior of BBO. This suggests practical guidance on configuring the inner unit horizon to enhance training efficiency and model performance.

NeurIPS Conference 2024 Conference Paper

Decomposed Prompt Decision Transformer for Efficient Unseen Task Generalization

• Hongling Zheng
• Li Shen
• Yong Luo
• Tongliang Liu
• Jialie Shen
• Dacheng Tao

Multi-task offline reinforcement learning aims to develop a unified policy for diverse tasks without requiring real-time interaction with the environment. Recent work explores sequence modeling, leveraging the scalability of the transformer architecture as a foundation for multi-task learning. Given the variations in task content and complexity, formulating policies becomes a challenging endeavor, requiring careful parameter sharing and adept management of conflicting gradients to extract rich cross-task knowledge from multiple tasks and transfer it to unseen tasks. In this paper, we propose the Decomposed Prompt Decision Transformer (DPDT) that adopts a two-stage paradigm to efficiently learn prompts for unseen tasks in a parameter-efficient manner. We incorporate parameters from pre-trained language models (PLMs) to initialize DPDT, thereby providing rich prior knowledge encoded in language models. During the decomposed prompt tuning phase, we learn both cross-task and task-specific prompts on training tasks to achieve prompt decomposition. In the test time adaptation phase, the cross-task prompt, serving as a good initialization, were further optimized on unseen tasks through test time adaptation, enhancing the model's performance on these tasks. Empirical evaluation on a series of Meta-RL benchmarks demonstrates the superiority of our approach. The project is available at https: //github. com/ruthless-man/DPDT.

NeurIPS Conference 2024 Conference Paper

Deep Graph Mating

• Yongcheng Jing
• Seok-Hee Hong
• Dacheng Tao

In this paper, we introduce the first learning-free model reuse task within the non-Euclidean domain, termed as Deep Graph Mating (Grama). We strive to create a child Graph Neural Network (GNN) that integrates knowledge from pre-trained parent models without requiring re-training, fine-tuning, or annotated labels. To this end, we begin by investigating the permutation invariance property of GNNs, which leads us to develop two vanilla approaches for Grama: Vanilla Parameter Interpolation (VPI) and Vanilla Alignment Prior to Interpolation (VAPI), both employing topology-independent interpolation in the parameter space. However, neither approach has achieved the anticipated results. Through theoretical analysis of VPI and VAPI, we identify critical challenges unique to Grama, including increased sensitivity to parameter misalignment and further the inherent topology-dependent complexities. Motivated by these findings, we propose the Dual-Message Coordination and Calibration (DuMCC) methodology, comprising the Parent Message Coordination (PMC) scheme to optimise the permutation matrices for parameter interpolation by coordinating aggregated messages, and the Child Message Calibration (CMC) scheme to mitigate over-smoothing identified in PMC by calibrating the message statistics within child GNNs. Experiments across diverse domains, including node and graph property prediction, 3D object recognition, and large-scale semantic parsing, demonstrate that the proposed DuMCC effectively enables training-free knowledge transfer, yielding results on par with those of pre-trained models.

ECAI Conference 2024 Conference Paper

Diversifying the Mixture-of-Experts Representation for Language Models with Orthogonal Optimizer

• Boan Liu
• Liang Ding 0006
• Li Shen 0008
• Keqin Peng
• Yu Cao 0014
• Dazhao Cheng
• Dacheng Tao

The Mixture of Experts (MoE) has emerged as a highly successful technique in deep learning, based on the principle of divide-and-conquer to maximize model capacity without significant additional computational cost. Even in the era of large-scale language models (LLMs), MoE continues to play a crucial role, as some researchers have indicated that GPT-4 adopts the MoE structure to ensure diverse inference results. However, MoE is susceptible to performance degeneracy, particularly evident in the issues of imbalance and homogeneous representation among experts. While previous studies have extensively addressed the problem of imbalance, the challenge of homogeneous representation remains unresolved. In this study, we shed light on the homogeneous representation problem, wherein experts in the MoE fail to specialize and lack diversity, leading to frustratingly high similarities in their representations (up to 99% in a well-performed MoE model). This problem restricts the expressive power of the MoE and, we argue, contradicts its original intention. To tackle this issue, we propose a straightforward yet highly effective solution: OMoE, an orthogonal expert optimizer. Additionally, we introduce an alternating training strategy that encourages each expert to update in a direction orthogonal to the subspace spanned by other experts. Our algorithm facilitates MoE training in two key ways: firstly, it explicitly enhances representation diversity, and secondly, it implicitly fosters interaction between experts during orthogonal weights computation. Through extensive experiments, we demonstrate that our proposed optimization algorithm significantly improves the performance of fine-tuning the MoE model on the GLUE benchmark, SuperGLUE benchmark, question-answering task, and name entity recognition tasks.

ICML Conference 2024 Conference Paper

Generalization Analysis of Stochastic Weight Averaging with General Sampling

• Peng Wang
• Li Shen 0008
• Zerui Tao
• Shuaida He
• Dacheng Tao

Stochastic weight averaging (SWA) method has empirically proven its advantages compared to stochastic gradient descent (SGD). Despite it is widespread used, theoretical investigations have been limited, particularly in scenarios beyond the ideal setting of convex and sampling with replacement. However, non-convex cases and sampling without replacement are very practical in real-world applications. The main challenges under the above settings are two-folds: (i) All the historical gradient information introduced by SWA is considered, while the analysis of SGD using the tool of uniform stability requires only to bound the current gradient. (ii) The $(1+\alpha\beta)$-expansion property causes the boundary of each gradient step dependent on the previous step, making the boundary of each historical gradient in SWA nested and the theoretical analysis even harder. To address the theoretical challenges, we adopt mathematical induction to find a recursive representation that bounds the gradient at each step. Based on this, we establish stability bounds supporting sampling with and without replacement in the non-convex setting. Furthermore, the derived generalization bounds of SWA are sharper than SGD. At last, experimental results on several benchmarks verify our theoretical results.

NeurIPS Conference 2024 Conference Paper

GoMatching: A Simple Baseline for Video Text Spotting via Long and Short Term Matching

• Haibin He
• Maoyuan Ye
• Jing Zhang
• Juhua Liu
• Bo Du
• Dacheng Tao

Beyond the text detection and recognition tasks in image text spotting, video text spotting presents an augmented challenge with the inclusion of tracking. While advanced end-to-end trainable methods have shown commendable performance, the pursuit of multi-task optimization may pose the risk of producing sub-optimal outcomes for individual tasks. In this paper, we identify a main bottleneck in the state-of-the-art video text spotter: the limited recognition capability. In response to this issue, we propose to efficiently turn an off-the-shelf query-based image text spotter into a specialist on video and present a simple baseline termed GoMatching, which focuses the training efforts on tracking while maintaining strong recognition performance. To adapt the image text spotter to video datasets, we add a rescoring head to rescore each detected instance's confidence via efficient tuning, leading to a better tracking candidate pool. Additionally, we design a long-short term matching module, termed LST-Matcher, to enhance the spotter's tracking capability by integrating both long- and short-term matching results via Transformer. Based on the above simple designs, GoMatching delivers new records on ICDAR15-video, DSText, BOVText, and our proposed novel test set with arbitrary-shaped text termed ArTVideo, which demonstates GoMatching's capability to accommodate general, dense, small, arbitrary-shaped, Chinese and English text scenarios while saving considerable training budgets. The code will be released.

ICML Conference 2024 Conference Paper

HarmoDT: Harmony Multi-Task Decision Transformer for Offline Reinforcement Learning

• Shengchao Hu
• Ziqing Fan
• Li Shen 0008
• Ya Zhang 0002
• Yanfeng Wang 0001
• Dacheng Tao

The purpose of offline multi-task reinforcement learning (MTRL) is to develop a unified policy applicable to diverse tasks without the need for online environmental interaction. Recent advancements approach this through sequence modeling, leveraging the Transformer architecture’s scalability and the benefits of parameter sharing to exploit task similarities. However, variations in task content and complexity pose significant challenges in policy formulation, necessitating judicious parameter sharing and management of conflicting gradients for optimal policy performance. In this work, we introduce the Harmony Multi-Task Decision Transformer (HarmoDT), a novel solution designed to identify an optimal harmony subspace of parameters for each task. We approach this as a bi-level optimization problem, employing a meta-learning framework that leverages gradient-based techniques. The upper level of this framework is dedicated to learning a task-specific mask that delineates the harmony subspace, while the inner level focuses on updating parameters to enhance the overall performance of the unified policy. Empirical evaluations on a series of benchmarks demonstrate the superiority of HarmoDT, verifying the effectiveness of our approach.

JMLR Journal 2024 Journal Article

Identifiability and Asymptotics in Learning Homogeneous Linear ODE Systems from Discrete Observations

• Yuanyuan Wang
• Wei Huang
• Mingming Gong
• Xi Geng
• Tongliang Liu
• Kun Zhang
• Dacheng Tao

Ordinary Differential Equations (ODEs) have recently gained a lot of attention in machine learning. However, the theoretical aspects, for example, identifiability and asymptotic properties of statistical estimation are still obscure. This paper derives a sufficient condition for the identifiability of homogeneous linear ODE systems from a sequence of equally-spaced error-free observations sampled from a single trajectory. When observations are disturbed by measurement noise, we prove that under mild conditions, the parameter estimator based on the Nonlinear Least Squares (NLS) method is consistent and asymptotic normal with $n^{-1/2}$ convergence rate. Based on the asymptotic normality property, we construct confidence sets for the unknown system parameters and propose a new method to infer the causal structure of the ODE system, that is, inferring whether there is a causal link between system variables. Furthermore, we extend the results to degraded observations, including aggregated and time-scaled ones. To the best of our knowledge, our work is the first systematic study of the identifiability and asymptotic properties in learning linear ODE systems. We also construct simulations with various system dimensions to illustrate the established theoretical results. [abs] [ pdf ][ bib ] &copy JMLR 2024. ( edit, beta )

NeurIPS Conference 2024 Conference Paper

InfoRM: Mitigating Reward Hacking in RLHF via Information-Theoretic Reward Modeling

• Yuchun Miao
• Sen Zhang
• Liang Ding
• Rong Bao
• Lefei Zhang
• Dacheng Tao

Despite the success of reinforcement learning from human feedback (RLHF) in aligning language models with human values, reward hacking, also termed reward overoptimization, remains a critical challenge. This issue primarily arises from reward misgeneralization, where reward models (RMs) compute reward using spurious features that are irrelevant to human preferences. In this work, we tackle this problem from an information-theoretic perspective and propose a framework for reward modeling, namely InfoRM, by introducing a variational information bottleneck objective to filter out irrelevant information. Notably, we further identify a correlation between overoptimization and outliers in the IB latent space of InfoRM, establishing it as a promising tool for detecting reward overoptimization. Inspired by this finding, we propose the Cluster Separation Index (CSI), which quantifies deviations in the IB latent space, as an indicator of reward overoptimization to facilitate the development of online mitigation strategies. Extensive experiments on a wide range of settings and RM scales (70M, 440M, 1. 4B, and 7B) demonstrate the effectiveness of InfoRM. Further analyses reveal that InfoRM's overoptimization detection mechanism is not only effective but also robust across a broad range of datasets, signifying a notable advancement in the field of RLHF. The code will be released upon acceptance.

IJCAI Conference 2024 Conference Paper

Joint Input and Output Coordination for Class-Incremental Learning

• Shuai Wang
• Yibing Zhan
• Yong Luo
• Han Hu
• Wei Yu
• Yonggang Wen
• Dacheng Tao

Incremental learning is nontrivial due to severe catastrophic forgetting. Although storing a small amount of data on old tasks during incremental learning is a feasible solution, current strategies still do not 1) adequately address the class bias problem, and 2) alleviate the mutual interference between new and old tasks, and 3) consider the problem of class bias within tasks. In light of the above issues, we analyze the cause of class bias in incremental learning, as well as the drawbacks of existing approaches, and propose a joint input and output coordination (JIOC) mechanism to address these issues. This mechanism assigns different weights to different categories of data according to the gradient of the output score, and uses knowledge distillation (KD) to reduce the mutual interference between the outputs of old and new tasks. The proposed mechanism is general and flexible, and can be incorporated into different incremental learning approaches that use memory storage. Extensive experiments show that our mechanism can significantly improve their performance.

ICLR Conference 2024 Conference Paper

Learning Multi-Agent Communication from Graph Modeling Perspective

• Shengchao Hu
• Li Shen 0008
• Ya Zhang 0002
• Dacheng Tao

In numerous artificial intelligence applications, the collaborative efforts of multiple intelligent agents are imperative for the successful attainment of target objectives. To enhance coordination among these agents, a distributed communication framework is often employed. However, information sharing among all agents proves to be resource-intensive, while the adoption of a manually pre-defined communication architecture imposes limitations on inter-agent communication, thereby constraining the potential for collaborative efforts. In this study, we introduce a novel approach wherein we conceptualize the communication architecture among agents as a learnable graph. We formulate this problem as the task of determining the communication graph while enabling the architecture parameters to update normally, thus necessitating a bi-level optimization process. Utilizing continuous relaxation of the graph representation and incorporating attention units, our proposed approach, CommFormer, efficiently optimizes the communication graph and concurrently refines architectural parameters through gradient descent in an end-to-end manner. Extensive experiments on a variety of cooperative tasks substantiate the robustness of our model across diverse cooperative scenarios, where agents are able to develop more coordinated and sophisticated strategies regardless of changes in the number of agents.

ICML Conference 2024 Conference Paper

Merging Multi-Task Models via Weight-Ensembling Mixture of Experts

• Anke Tang
• Li Shen 0008
• Yong Luo 0002
• Nan Yin
• Lefei Zhang
• Dacheng Tao

Merging various task-specific Transformer-based vision models trained on different tasks into a single unified model can execute all the tasks concurrently. Previous methods, exemplified by task arithmetic, have been proven to be both effective and scalable. Existing methods have primarily focused on seeking a static optimal solution within the original model parameter space. A notable challenge is mitigating the interference between parameters of different models, which can substantially deteriorate performance. In this paper, we propose to merge most of the parameters while upscaling the MLP of the Transformer layers to a weight-ensembling mixture of experts (MoE) module, which can dynamically integrate shared and task-specific knowledge based on the input, thereby providing a more flexible solution that can adapt to the specific needs of each instance. Our key insight is that by identifying and separating shared knowledge and task-specific knowledge, and then dynamically integrating them, we can mitigate the parameter interference problem to a great extent. We conduct the conventional multi-task model merging experiments and evaluate the generalization and robustness of our method. The results demonstrate the effectiveness of our method and provide a comprehensive understanding of our method. The code is available at https: //github. com/tanganke/weight-ensembling_MoE

NeurIPS Conference 2024 Conference Paper

MG-Net: Learn to Customize QAOA with Circuit Depth Awareness

• Yang Qian
• Xinbiao Wang
• Yuxuan Du
• Yong Luo
• Dacheng Tao

Quantum Approximate Optimization Algorithm (QAOA) and its variants exhibit immense potential in tackling combinatorial optimization challenges. However, their practical realization confronts a dilemma: the requisite circuit depth for satisfactory performance is problem-specific and often exceeds the maximum capability of current quantum devices. To address this dilemma, here we first analyze the convergence behavior of QAOA, uncovering the origins of this dilemma and elucidating the intricate relationship between the employed mixer Hamiltonian, the specific problem at hand, and the permissible maximum circuit depth. Harnessing this understanding, we introduce the Mixer Generator Network (MG-Net), a unified deep learning framework adept at dynamically formulating optimal mixer Hamiltonians tailored to distinct tasks and circuit depths. Systematic simulations, encompassing Ising models and weighted Max-Cut instances with up to 64 qubits, substantiate our theoretical findings, highlighting MG-Net's superior performance in terms of both approximation ratio and efficiency.

AAAI Conference 2024 Conference Paper

Multi-Step Denoising Scheduled Sampling: Towards Alleviating Exposure Bias for Diffusion Models

• Zhiyao Ren
• Yibing Zhan
• Liang Ding
• Gaoang Wang
• Chaoyue Wang
• Zhongyi Fan
• Dacheng Tao

Denoising Diffusion Probabilistic Models (DDPMs) have achieved significant success in generation tasks. Nevertheless, the exposure bias issue, i.e., the natural discrepancy between the training (the output of each step is calculated individually by a given input) and inference (the output of each step is calculated based on the input iteratively obtained based on the model), harms the performance of DDPMs. To our knowledge, few works have tried to tackle this issue by modifying the training process for DDPMs, but they still perform unsatisfactorily due to 1) partially modeling the discrepancy and 2) ignoring the prediction error accumulation. To address the above issues, in this paper, we propose a multi-step denoising scheduled sampling (MDSS) strategy to alleviate the exposure bias for DDPMs. Analyzing the formulations of the training and inference of DDPMs, MDSS 1) comprehensively considers the discrepancy influence of prediction errors on the output of the model (the Gaussian noise) and the output of the step (the calculated input signal of the next step), and 2) efficiently models the prediction error accumulation by using multiple iterations of a mathematical formulation initialized from one-step prediction error obtained from the model. The experimental results, compared with previous works, demonstrate that our approach is more effective in mitigating exposure bias in DDPM, DDIM, and DPM-solver. In particular, MDSS achieves an FID score of 3.86 in 100 sample steps of DDIM on the CIFAR-10 dataset, whereas the second best obtains 4.78. The code will be available on GitHub.

NeurIPS Conference 2024 Conference Paper

Offline Behavior Distillation

• Shiye Lei
• Sen Zhang
• Dacheng Tao

Massive reinforcement learning (RL) data are typically collected to train policies offline without the need for interactions, but the large data volume can cause training inefficiencies. To tackle this issue, we formulate offline behavior distillation (OBD), which synthesizes limited expert behavioral data from sub-optimal RL data, enabling rapid policy learning. We propose two naive OBD objectives, DBC and PBC, which measure distillation performance via the decision difference between policies trained on distilled data and either offline data or a near-expert policy. Due to intractable bi-level optimization, the OBD objective is difficult to minimize to small values, which deteriorates PBC by its distillation performance guarantee with quadratic discount complexity $\mathcal{O}(1/(1-\gamma)^2)$. We theoretically establish the equivalence between the policy performance and action-value weighted decision difference, and introduce action-value weighted PBC (Av-PBC) as a more effective OBD objective. By optimizing the weighted decision difference, Av-PBC achieves a superior distillation guarantee with linear discount complexity $\mathcal{O}(1/(1-\gamma))$. Extensive experiments on multiple D4RL datasets reveal that Av-PBC offers significant improvements in OBD performance, fast distillation convergence speed, and robust cross-architecture/optimizer generalization.

ICLR Conference 2024 Conference Paper

One For All: Towards Training One Graph Model For All Classification Tasks

• Hao Liu 0057
• Jiarui Feng
• Lecheng Kong
• Ningyue Liang
• Dacheng Tao
• Yixin Chen 0001
• Muhan Zhang

Designing a single model to address multiple tasks has been a long-standing objective in artificial intelligence. Recently, large language models have demonstrated exceptional capability in solving different tasks within the language domain. However, a unified model for various graph tasks remains underexplored, primarily due to the challenges unique to the graph learning domain. First, graph data from different areas carry distinct attributes and follow different distributions. Such discrepancy makes it hard to represent graphs in a single representation space. Second, tasks on graphs diversify into node, link, and graph tasks, requiring distinct embedding strategies. Finally, an appropriate graph prompting paradigm for in-context learning is unclear. We propose **One for All (OFA)**, the first general framework that can use a single graph model to address the above challenges. Specifically, OFA proposes text-attributed graphs to unify different graph data by describing nodes and edges with natural language and uses language models to encode the diverse and possibly cross-domain text attributes to feature vectors in the same embedding space. Furthermore, OFA introduces the concept of nodes-of-interest to standardize different tasks with a single task representation. For in-context learning on graphs, OFA introduces a novel graph prompting paradigm that appends prompting substructures to the input graph, which enables it to address varied tasks without fine-tuning. We train the OFA model using graph data from multiple domains (including citation networks, molecular graphs, knowledge graphs, etc.) simultaneously and evaluate its ability in supervised, few-shot, and zero-shot learning scenarios. OFA performs well across different tasks, making it the first general-purpose across-domains classification model on graphs.

ICLR Conference 2024 Conference Paper

Parameter-Efficient Multi-Task Model Fusion with Partial Linearization

• Anke Tang
• Li Shen 0008
• Yong Luo 0002
• Yibing Zhan
• Han Hu 0003
• Bo Du 0001
• Yixin Chen 0001
• Dacheng Tao

Large pre-trained models have enabled significant advances in machine learning and served as foundation components. Model fusion methods, such as task arithmetic, have been proven to be powerful and scalable to incorporate fine-tuned weights from different tasks into a multi-task model. However, efficiently fine-tuning large pre-trained models on multiple downstream tasks remains challenging, leading to inefficient multi-task model fusion. In this work, we propose a novel method to improve multi-task fusion for parameter-efficient fine-tuning techniques like LoRA fine-tuning. Specifically, our approach partially linearizes only the adapter modules and applies task arithmetic over the linearized adapters. This allows us to leverage the the advantages of model fusion over linearized fine-tuning, while still performing fine-tuning and inference efficiently. We demonstrate that our partial linearization technique enables a more effective fusion of multiple tasks into a single model, outperforming standard adapter tuning and task arithmetic alone. Experimental results demonstrate the capabilities of our proposed partial linearization technique to effectively construct unified multi-task models via the fusion of fine-tuned task vectors. We evaluate performance over an increasing number of tasks and find that our approach outperforms standard parameter-efficient fine-tuning techniques. The results highlight the benefits of partial linearization for scalable and efficient multi-task model fusion.

ICML Conference 2024 Conference Paper

Q-value Regularized Transformer for Offline Reinforcement Learning

• Shengchao Hu
• Ziqing Fan
• Chaoqin Huang
• Li Shen 0008
• Ya Zhang 0002
• Yanfeng Wang 0001
• Dacheng Tao

Recent advancements in offline reinforcement learning (RL) have underscored the capabilities of Conditional Sequence Modeling (CSM), a paradigm that learns the action distribution based on history trajectory and target returns for each state. However, these methods often struggle with stitching together optimal trajectories from sub-optimal ones due to the inconsistency between the sampled returns within individual trajectories and the optimal returns across multiple trajectories. Fortunately, Dynamic Programming (DP) methods offer a solution by leveraging a value function to approximate optimal future returns for each state, while these techniques are prone to unstable learning behaviors, particularly in long-horizon and sparse-reward scenarios. Building upon these insights, we propose the Q-value regularized Transformer (QT), which combines the trajectory modeling ability of the Transformer with the predictability of optimal future returns from DP methods. QT learns an action-value function and integrates a term maximizing action-values into the training loss of CSM, which aims to seek optimal actions that align closely with the behavior policy. Empirical evaluations on D4RL benchmark datasets demonstrate the superiority of QT over traditional DP and CSM methods, highlighting the potential of QT to enhance the state-of-the-art in offline RL.

ICML Conference 2024 Conference Paper

Representation Surgery for Multi-Task Model Merging

• Enneng Yang
• Li Shen 0008
• Zhenyi Wang 0001
• Guibing Guo
• Xiaojun Chen 0006
• Xingwei Wang 0001
• Dacheng Tao

Multi-task learning (MTL) compresses the information from multiple tasks into a unified backbone to improve computational efficiency and generalization. Recent work directly merges multiple independently trained models to perform MTL instead of collecting their raw data for joint training, greatly expanding the application scenarios of MTL. However, by visualizing the representation distribution of existing model merging schemes, we find that the merged model often suffers from the dilemma of representation bias. That is, there is a significant discrepancy in the representation distribution between the merged and individual models, resulting in poor performance of merged MTL. In this paper, we propose a representation surgery solution called “Surgery" to reduce representation bias in the merged model. Specifically, Surgery is a lightweight task-specific plugin that takes the representation of the merged model as input and attempts to output the biases contained in the representation from the merged model. We then designed an unsupervised optimization objective that updates the Surgery plugin by minimizing the distance between the merged model’s representation and the individual model’s representation. Extensive experiments demonstrate significant MTL performance improvements when our Surgery plugin is applied to state-of-the-art (SOTA) model merging schemes.

ICLR Conference 2024 Conference Paper

Revisiting Plasticity in Visual Reinforcement Learning: Data, Modules and Training Stages

• Guozheng Ma
• Lu Li
• Sen Zhang 0006
• Zixuan Liu 0002
• Zhen Wang 0030
• Yixin Chen 0001
• Li Shen 0008
• Xueqian Wang 0001

Plasticity, the ability of a neural network to evolve with new data, is crucial for high-performance and sample-efficient visual reinforcement learning (VRL). Although methods like resetting and regularization can potentially mitigate plasticity loss, the influences of various components within the VRL framework on the agent's plasticity are still poorly understood. In this work, we conduct a systematic empirical exploration focusing on three primary underexplored facets and derive the following insightful conclusions: (1) data augmentation is essential in maintaining plasticity; (2) the critic's plasticity loss serves as the principal bottleneck impeding efficient training; and (3) without timely intervention to recover critic's plasticity in the early stages, its loss becomes catastrophic. These insights suggest a novel strategy to address the high replay ratio (RR) dilemma, where exacerbated plasticity loss hinders the potential improvements of sample efficiency brought by increased reuse frequency. Rather than setting a static RR for the entire training process, we propose Adaptive RR, which dynamically adjusts the RR based on the critic’s plasticity level. Extensive evaluations indicate that Adaptive RR not only avoids catastrophic plasticity loss in the early stages but also benefits from more frequent reuse in later phases, resulting in superior sample efficiency.

ICRA Conference 2024 Conference Paper

Semantic-focused Patch Tokenizer with Multi-branch Mixer for Visual Place Recognition

• Zhenyu Xu
• Ziliang Ren
• Qieshi Zhang
• Jie Lou
• Dacheng Tao
• Jun Cheng 0002

Visual Place Recognition (VPR) is critical for navigation and loop closure in autonomous driving tasks, mitigating the impact of shift errors caused by dynamic changes in the environment. Due to the limited ability of backbone networks and extreme environmental changes, current methods fail to capture foundational semantic details that include the distinctive attributes for unique place identification. To address this problem, we propose a new visual token-guided VPR framework that contains a semantic-focused patch tokenizer and a multi-branch Mixer. To mitigate the inference from place-unrelated objects, the semantic-focused patch tokenizer exploits attention-based channel selection and spatial partition, which efficiently captures important semantic information within the channels and preserve spatial relationships among the backbone features. To extract abstract features with spatial structure information, the multi-branch Mixer utilizes a multi-branch structure to aggregate local and global position information, improving the robustness of global representations to environmental changes. Experimental results demonstrate that our method outperforms state-of-the-art methods, achieving 85. 3% Recall@1 on the MSLS val dataset and 59. 1% Recall@1 on the Nordland dataset when using ResNet18 as the backbone.

AAAI Conference 2024 Conference Paper

SimDistill: Simulated Multi-Modal Distillation for BEV 3D Object Detection

• Haimei Zhao
• Qiming Zhang
• Shanshan Zhao
• Zhe Chen
• Jing Zhang
• Dacheng Tao

Multi-view camera-based 3D object detection has become popular due to its low cost, but accurately inferring 3D geometry solely from camera data remains challenging and may lead to inferior performance. Although distilling precise 3D geometry knowledge from LiDAR data could help tackle this challenge, the benefits of LiDAR information could be greatly hindered by the significant modality gap between different sensory modalities. To address this issue, we propose a Simulated multi-modal Distillation (SimDistill) method by carefully crafting the model architecture and distillation strategy. Specifically, we devise multi-modal architectures for both teacher and student models, including a LiDAR-camera fusion-based teacher and a simulated fusion-based student. Owing to the ``identical'' architecture design, the student can mimic the teacher to generate multi-modal features with merely multi-view images as input, where a geometry compensation module is introduced to bridge the modality gap. Furthermore, we propose a comprehensive multi-modal distillation scheme that supports intra-modal, cross-modal, and multi-modal fusion distillation simultaneously in the Bird's-eye-view space. Incorporating them together, our SimDistill can learn better feature representations for 3D object detection while maintaining a cost-effective camera-only deployment. Extensive experiments validate the effectiveness and superiority of SimDistill over state-of-the-art methods, achieving an improvement of 4.8% mAP and 4.1% NDS over the baseline detector. The source code will be released at https://github.com/ViTAE-Transformer/SimDistill.

ICML Conference 2024 Conference Paper

Sparse Model Inversion: Efficient Inversion of Vision Transformers for Data-Free Applications

• Zixuan Hu
• Yongxian Wei
• Li Shen 0008
• Zhenyi Wang 0001
• Lei Li 0051
• Chun Yuan 0003
• Dacheng Tao

Model inversion, which aims to reconstruct the original training data from pre-trained discriminative models, is especially useful when the original training data is unavailable due to privacy, usage rights, or size constraints. However, existing dense inversion methods attempt to reconstruct the entire image area, making them extremely inefficient when inverting high-resolution images from large-scale Vision Transformers (ViTs). We further identify two underlying causes of this inefficiency: the redundant inversion of noisy backgrounds and the unintended inversion of spurious correlations—a phenomenon we term “hallucination” in model inversion. To address these limitations, we propose a novel sparse model inversion strategy, as a plug-and-play extension to speed up existing dense inversion methods with no need for modifying their original loss functions. Specifically, we selectively invert semantic foregrounds while stopping the inversion of noisy backgrounds and potential spurious correlations. Through both theoretical and empirical studies, we validate the efficacy of our approach in achieving significant inversion acceleration (up to $\times$3. 79) while maintaining comparable or even enhanced downstream performance in data-free model quantization and data-free knowledge transfer. Code is available at https: //github. com/Egg-Hu/SMI.

ICML Conference 2024 Conference Paper

Task Groupings Regularization: Data-Free Meta-Learning with Heterogeneous Pre-trained Models

• Yongxian Wei
• Zixuan Hu
• Li Shen 0008
• Zhenyi Wang 0001
• Yu Li 0006
• Chun Yuan 0003
• Dacheng Tao

Data-Free Meta-Learning (DFML) aims to derive knowledge from a collection of pre-trained models without accessing their original data, enabling the rapid adaptation to new unseen tasks. Current methods often overlook the heterogeneity among pre-trained models, which leads to performance degradation due to task conflicts. In this paper, we empirically and theoretically identify and analyze the model heterogeneity in DFML. We find that model heterogeneity introduces a heterogeneity-homogeneity trade-off, where homogeneous models reduce task conflicts but also increase the overfitting risk. Balancing this trade-off is crucial for learning shared representations across tasks. Based on our findings, we propose Task Groupings Regularization, a novel approach that benefits from model heterogeneity by grouping and aligning conflicting tasks. Specifically, we embed pre-trained models into a task space to compute dissimilarity, and group heterogeneous models together based on this measure. Then, we introduce implicit gradient regularization within each group to mitigate potential conflicts. By encouraging a gradient direction suitable for all tasks, the meta-model captures shared representations that generalize across tasks. Comprehensive experiments showcase the superiority of our approach in multiple benchmarks, effectively tackling the model heterogeneity in challenging multi-domain and multi-architecture scenarios.

AAAI Conference 2024 Conference Paper

TD²-Net: Toward Denoising and Debiasing for Video Scene Graph Generation

• Xin Lin
• Chong Shi
• Yibing Zhan
• Zuopeng Yang
• Yaqi Wu
• Dacheng Tao

Dynamic scene graph generation (SGG) focuses on detecting objects in a video and determining their pairwise relationships. Existing dynamic SGG methods usually suffer from several issues, including 1) Contextual noise, as some frames might contain occluded and blurred objects. 2) Label bias, primarily due to the high imbalance between a few positive relationship samples and numerous negative ones. Additionally, the distribution of relationships exhibits a long-tailed pattern. To address the above problems, in this paper, we introduce a network named TD2-Net that aims at denoising and debiasing for dynamic SGG. Specifically, we first propose a denoising spatio-temporal transformer module that enhances object representation with robust contextual information. This is achieved by designing a differentiable Top-K object selector that utilizes the gumbel-softmax sampling strategy to select the relevant neighborhood for each object. Second, we introduce an asymmetrical reweighting loss to relieve the issue of label bias. This loss function integrates asymmetry focusing factors and the volume of samples to adjust the weights assigned to individual samples. Systematic experimental results demonstrate the superiority of our proposed TD2-Net over existing state-of-the-art approaches on Action Genome databases. In more detail, TD2-Net outperforms the second-best competitors by 12.7% on mean-Recall@10 for predicate classification.

ICML Conference 2024 Conference Paper

Towards Theoretical Understandings of Self-Consuming Generative Models

• Shi Fu
• Sen Zhang 0006
• Yingjie Wang 0007
• Xinmei Tian 0001
• Dacheng Tao

This paper tackles the emerging challenge of training generative models within a self-consuming loop, wherein successive generations of models are recursively trained on mixtures of real and synthetic data from previous generations. We construct a theoretical framework to rigorously evaluate how this training procedure impacts the data distributions learned by future models, including parametric and non-parametric models. Specifically, we derive bounds on the total variation (TV) distance between the synthetic data distributions produced by future models and the original real data distribution under various mixed training scenarios for diffusion models with a one-hidden-layer neural network score function. Our analysis demonstrates that this distance can be effectively controlled under the condition that mixed training dataset sizes or proportions of real data are large enough. Interestingly, we further unveil a phase transition induced by expanding synthetic data amounts, proving theoretically that while the TV distance exhibits an initial ascent, it declines beyond a threshold point. Finally, we present results for kernel density estimation, delivering nuanced insights such as the impact of mixed data training on error propagation.

TMLR Journal 2024 Journal Article

Visual Prompt Based Personalized Federated Learning

• Guanghao Li
• Wansen Wu
• Yan Sun
• Li Shen
• Baoyuan Wu
• Dacheng Tao

As a popular paradigm of distributed learning, personalized federated learning (PFL) allows personalized models to improve generalization ability and robustness by utilizing knowledge from all distributed clients. Most existing PFL algorithms tackle personalization in a model-centric way, such as personalized layer partition, model regularization, and model interpolation, which all fail to take into account the data characteristics of distributed clients. In this paper, we propose a novel PFL framework for image classification tasks, dubbed pFedPT, that leverages personalized visual prompts to implicitly represent local data distribution information of clients and provides that information to the aggregation model to help with classification tasks. Specifically, in each round of pFedPT training, each client generates a local personalized prompt related to local data distribution. Then, the local model is trained on the input composed of raw data and a visual prompt to learn the distribution information contained in the prompt. During model testing, the aggregated model obtains client-specific knowledge of the data distributions based on the prompts, which can be seen as an adaptive fine-tuning of the aggregation model to improve model performances on different clients. Furthermore, the visual prompt can be added as an orthogonal method to implement personalization on the client for existing FL methods to boost their performance. Experiments on the CIFAR10 and CIFAR100 datasets show that pFedPT outperforms several state-of-the-art (SOTA) PFL algorithms by a large margin in various settings. The code is available at: https://github.com/hkgdifyu/pFedPT.

NeurIPS Conference 2023 Conference Paper

All Points Matter: Entropy-Regularized Distribution Alignment for Weakly-supervised 3D Segmentation

• Liyao Tang
• Zhe Chen
• Shanshan Zhao
• Chaoyue Wang
• Dacheng Tao

Pseudo-labels are widely employed in weakly supervised 3D segmentation tasks where only sparse ground-truth labels are available for learning. Existing methods often rely on empirical label selection strategies, such as confidence thresholding, to generate beneficial pseudo-labels for model training. This approach may, however, hinder the comprehensive exploitation of unlabeled data points. We hypothesize that this selective usage arises from the noise in pseudo-labels generated on unlabeled data. The noise in pseudo-labels may result in significant discrepancies between pseudo-labels and model predictions, thus confusing and affecting the model training greatly. To address this issue, we propose a novel learning strategy to regularize the generated pseudo-labels and effectively narrow the gaps between pseudo-labels and model predictions. More specifically, our method introduces an Entropy Regularization loss and a Distribution Alignment loss for weakly supervised learning in 3D segmentation tasks, resulting in an ERDA learning strategy. Interestingly, by using KL distance to formulate the distribution alignment loss, it reduces to a deceptively simple cross-entropy-based loss which optimizes both the pseudo-label generation network and the 3D segmentation network simultaneously. Despite the simplicity, our method promisingly improves the performance. We validate the effectiveness through extensive experiments on various baselines and large-scale datasets. Results show that ERDA effectively enables the effective usage of all unlabeled data points for learning and achieves state-of-the-art performance under different settings. Remarkably, our method can outperform fully-supervised baselines using only 1\% of true annotations. Code and model will be made publicly available at https: //github. com/LiyaoTang/ERDA.

ICLR Conference 2023 Conference Paper

Alternating Differentiation for Optimization Layers

• Haixiang Sun
• Ye Shi 0001
• Jingya Wang 0001
• Hoang Duong Tuan
• H. Vincent Poor
• Dacheng Tao

The idea of embedding optimization problems into deep neural networks as optimization layers to encode constraints and inductive priors has taken hold in recent years. Most existing methods focus on implicitly differentiating Karush–Kuhn–Tucker (KKT) conditions in a way that requires expensive computations on the Jacobian matrix, which can be slow and memory-intensive. In this paper, we developed a new framework, named Alternating Differentiation (Alt-Diff), that differentiates optimization problems (here, specifically in the form of convex optimization problems with polyhedral constraints) in a fast and recursive way. Alt-Diff decouples the differentiation procedure into a primal update and a dual update in an alternating way. Accordingly, Alt-Diff substantially decreases the dimensions of the Jacobian matrix especially for optimization with large-scale constraints and thus increases the computational speed of implicit differentiation. We show that the gradients obtained by Alt-Diff are consistent with those obtained by differentiating KKT conditions. In addition, we propose to truncate Alt-Diff to further accelerate the computational speed. Under some standard assumptions, we show that the truncation error of gradients is upper bounded by the same order of variables' estimation error. Therefore, Alt-Diff can be truncated to further increase computational speed without sacrificing much accuracy. A series of comprehensive experiments validate the superiority of Alt-Diff.

ICLR Conference 2023 Conference Paper

CktGNN: Circuit Graph Neural Network for Electronic Design Automation

• Zehao Dong
• Weidong Cao 0001
• Muhan Zhang
• Dacheng Tao
• Yixin Chen 0001
• Xuan Zhang 0001

The electronic design automation of analog circuits has been a longstanding challenge in the integrated circuit field due to the huge design space and complex design trade-offs among circuit specifications. In the past decades, intensive research efforts have only been paid to automate the transistor sizing with a given circuit topology. By recognizing the graph nature of circuits, this paper presents a Circuit Graph Neural Network (CktGNN) that simultaneously automates the circuit topology generation and device sizing based on the encoder-dependent optimization subroutines. Particularly, CktGNN encodes circuit graphs using a two-level GNN framework (of nested GNN) where circuits are represented as combinations of subgraphs in a known subgraph basis. In this way, it significantly improves efficiency by reducing the number of subgraphs to perform message passing. Nonetheless, another critical roadblock to advancing learning-assisted circuit design automation is a lack of public benchmarks to perform canonical assessment and reproducible research. To tackle the challenge, we introduce Open Circuit Benchmark (OCB), an open-sourced dataset that contains $10$K distinct operational amplifiers with carefully-extracted circuit specifications from physical implementations. OCB also equips with communicative circuit generation and evaluation capabilities such that it can be used to generalize the applicability of CktGNN to design various analog circuits by efficiently producing corresponding datasets. Experiments on OCB show the extraordinary advantages of CktGNN through representation-based optimization frameworks over other recent powerful GNN baselines and manual design from human experts. Our work paves the way toward a learning-based open-sourced design automation flow for analog circuits.

NeurIPS Conference 2023 Conference Paper

Cocktail: Mixing Multi-Modality Control for Text-Conditional Image Generation

• Minghui Hu
• Jianbin Zheng
• Daqing Liu
• Chuanxia Zheng
• Chaoyue Wang
• Dacheng Tao
• Tat-Jen Cham

Text-conditional diffusion models are able to generate high-fidelity images with diverse contents. However, linguistic representations frequently exhibit ambiguous descriptions of the envisioned objective imagery, requiring the incorporation of additional control signals to bolster the efficacy of text-guided diffusion models. In this work, we propose Cocktail, a pipeline to mix various modalities into one embedding, amalgamated with a generalized ControlNet (gControlNet), a controllable normalisation (ControlNorm), and a spatial guidance sampling method, to actualize multi-modal and spatially-refined control for text-conditional diffusion models. Specifically, we introduce a hyper-network gControlNet, dedicated to the alignment and infusion of the control signals from disparate modalities into the pre-trained diffusion model. gControlNet is capable of accepting flexible modality signals, encompassing the simultaneous reception of any combination of modality signals, or the supplementary fusion of multiple modality signals. The control signals are then fused and injected into the backbone model according to our proposed ControlNorm. Furthermore, our advanced spatial guidance sampling methodology proficiently incorporates the control signal into the designated region, thereby circumventing the manifestation of undesired objects within the generated image. We demonstrate the results of our method in controlling various modalities, proving high-quality synthesis and fidelity to multiple external signals.

NeurIPS Conference 2023 Conference Paper

ConDaFormer: Disassembled Transformer with Local Structure Enhancement for 3D Point Cloud Understanding

• Lunhao Duan
• Shanshan Zhao
• Nan Xue
• Mingming Gong
• Gui-Song Xia
• Dacheng Tao

Transformers have been recently explored for 3D point cloud understanding with impressive progress achieved. A large number of points, over 0. 1 million, make the global self-attention infeasible for point cloud data. Thus, most methods propose to apply the transformer in a local region, e. g. , spherical or cubic window. However, it still contains a large number of Query-Key pairs, which requires high computational costs. In addition, previous methods usually learn the query, key, and value using a linear projection without modeling the local 3D geometric structure. In this paper, we attempt to reduce the costs and model the local geometry prior by developing a new transformer block, named ConDaFormer. Technically, ConDaFormer disassembles the cubic window into three orthogonal 2D planes, leading to fewer points when modeling the attention in a similar range. The disassembling operation is beneficial to enlarging the range of attention without increasing the computational complexity, but ignores some contexts. To provide a remedy, we develop a local structure enhancement strategy that introduces a depth-wise convolution before and after the attention. This scheme can also capture the local geometric information. Taking advantage of these designs, ConDaFormer captures both long-range contextual information and local priors. The effectiveness is demonstrated by experimental results on several 3D point cloud understanding benchmarks. Our code will be available.

AAAI Conference 2023 Conference Paper

ContraFeat: Contrasting Deep Features for Semantic Discovery

• Xinqi Zhu
• Chang Xu
• Dacheng Tao

StyleGAN has shown strong potential for disentangled semantic control, thanks to its special design of multi-layer intermediate latent variables. However, existing semantic discovery methods on StyleGAN rely on manual selection of modified latent layers to obtain satisfactory manipulation results, which is tedious and demanding. In this paper, we propose a model that automates this process and achieves state-of-the-art semantic discovery performance. The model consists of an attention-equipped navigator module and losses contrasting deep-feature changes. We propose two model variants, with one contrasting samples in a binary manner, and another one contrasting samples with learned prototype variation patterns. The proposed losses are computed with pretrained deep features, based on our assumption that the features implicitly possess the desired semantic variation structure including consistency and orthogonality. Additionally, we design two metrics to quantitatively evaluate the performance of semantic discovery methods on FFHQ dataset, and also show that disentangled representations can be derived via a simple training process. Experimentally, we show that our models achieve state-of-the-art semantic discovery results without relying on layer-wise manual selection, and these discovered semantics can be used to manipulate real-world images.

ICML Conference 2023 Conference Paper

Decentralized SGD and Average-direction SAM are Asymptotically Equivalent

• Tongtian Zhu
• Fengxiang He
• Kaixuan Chen 0004
• Mingli Song
• Dacheng Tao

Decentralized stochastic gradient descent (D-SGD) allows collaborative learning on massive devices simultaneously without the control of a central server. However, existing theories claim that decentralization invariably undermines generalization. In this paper, we challenge the conventional belief and present a completely new perspective for understanding decentralized learning. We prove that D-SGD implicitly minimizes the loss function of an average-direction Sharpness-aware minimization (SAM) algorithm under general non-convex non-$\beta$-smooth settings. This surprising asymptotic equivalence reveals an intrinsic regularization-optimization trade-off and three advantages of decentralization: (1) there exists a free uncertainty evaluation mechanism in D-SGD to improve posterior estimation; (2) D-SGD exhibits a gradient smoothing effect; and (3) the sharpness regularization effect of D-SGD does not decrease as total batch size increases, which justifies the potential generalization benefit of D-SGD over centralized SGD (C-SGD) in large-batch scenarios.

NeurIPS Conference 2023 Conference Paper

Domain Re-Modulation for Few-Shot Generative Domain Adaptation

• Yi Wu
• Ziqiang Li
• Chaoyue Wang
• Heliang Zheng
• Shanshan Zhao
• Bin Li
• Dacheng Tao

In this study, we delve into the task of few-shot Generative Domain Adaptation (GDA), which involves transferring a pre-trained generator from one domain to a new domain using only a few reference images. Inspired by the way human brains acquire knowledge in new domains, we present an innovative generator structure called $\textbf{Domain Re-Modulation (DoRM)}$. DoRM not only meets the criteria of $\textit{high quality}$, $\textit{large synthesis diversity}$, and $\textit{cross-domain consistency}$, which were achieved by previous research in GDA, but also incorporates $\textit{memory}$ and $\textit{domain association}$, akin to how human brains operate. Specifically, DoRM freezes the source generator and introduces new mapping and affine modules (M\&A modules) to capture the attributes of the target domain during GDA. This process resembles the formation of new synapses in human brains. Consequently, a linearly combinable domain shift occurs in the style space. By incorporating multiple new M\&A modules, the generator gains the capability to perform high-fidelity multi-domain and hybrid-domain generation. Moreover, to maintain cross-domain consistency more effectively, we introduce a similarity-based structure loss. This loss aligns the auto-correlation map of the target image with its corresponding auto-correlation map of the source image during training. Through extensive experiments, we demonstrate the superior performance of our DoRM and similarity-based structure loss in few-shot GDA, both quantitatively and qualitatively. Code will be available at https: //github. com/wuyi2020/DoRM.

AAAI Conference 2023 Conference Paper

DPText-DETR: Towards Better Scene Text Detection with Dynamic Points in Transformer

• Maoyuan Ye
• Jing Zhang
• Shanshan Zhao
• Juhua Liu
• Bo Du
• Dacheng Tao

Recently, Transformer-based methods, which predict polygon points or Bezier curve control points for localizing texts, are popular in scene text detection. However, these methods built upon detection transformer framework might achieve sub-optimal training efficiency and performance due to coarse positional query modeling. In addition, the point label form exploited in previous works implies the reading order of humans, which impedes the detection robustness from our observation. To address these challenges, this paper proposes a concise Dynamic Point Text DEtection TRansformer network, termed DPText-DETR. In detail, DPText-DETR directly leverages explicit point coordinates to generate position queries and dynamically updates them in a progressive way. Moreover, to improve the spatial inductive bias of non-local self-attention in Transformer, we present an Enhanced Factorized Self-Attention module which provides point queries within each instance with circular shape guidance. Furthermore, we design a simple yet effective positional label form to tackle the side effect of the previous form. To further evaluate the impact of different label forms on the detection robustness in real-world scenario, we establish an Inverse-Text test set containing 500 manually labeled images. Extensive experiments prove the high training efficiency, robustness, and state-of-the-art performance of our method on popular benchmarks. The code and the Inverse-Text test set are available at https://github.com/ymy-k/DPText-DETR.

ICML Conference 2023 Conference Paper

Dynamic Regularized Sharpness Aware Minimization in Federated Learning: Approaching Global Consistency and Smooth Landscape

• Yan Sun
• Li Shen 0008
• Shixiang Chen
• Liang Ding 0006
• Dacheng Tao

In federated learning (FL), a cluster of local clients are chaired under the coordination of the global server and cooperatively train one model with privacy protection. Due to the multiple local updates and the isolated non-iid dataset, clients are prone to overfit into their own optima, which extremely deviates from the global objective and significantly undermines the performance. Most previous works only focus on enhancing the consistency between the local and global objectives to alleviate this prejudicial client drifts from the perspective of the optimization view, whose performance would be prominently deteriorated on the high heterogeneity. In this work, we propose a novel and general algorithm FedSMOO by jointly considering the optimization and generalization targets to efficiently improve the performance in FL. Concretely, FedSMOO adopts a dynamic regularizer to guarantee the local optima towards the global objective, which is meanwhile revised by the global Sharpness Aware Minimization (SAM) optimizer to search for the consistent flat minima. Our theoretical analysis indicates that FedSMOO achieves fast $\mathcal{O}(1/T)$ convergence rate with low generalization bound. Extensive numerical studies are conducted on the real-world dataset to verify its peerless efficiency and excellent generality.

AAAI Conference 2023 Conference Paper

Evaluating Model-Free Reinforcement Learning toward Safety-Critical Tasks

• Linrui Zhang
• Qin Zhang
• Li Shen
• Bo Yuan
• Xueqian Wang
• Dacheng Tao

Safety comes first in many real-world applications involving autonomous agents. Despite a large number of reinforcement learning (RL) methods focusing on safety-critical tasks, there is still a lack of high-quality evaluation of those algorithms that adheres to safety constraints at each decision step under complex and unknown dynamics. In this paper, we revisit prior work in this scope from the perspective of state-wise safe RL and categorize them as projection-based, recovery-based, and optimization-based approaches, respectively. Furthermore, we propose Unrolling Safety Layer (USL), a joint method that combines safety optimization and safety projection. This novel technique explicitly enforces hard constraints via the deep unrolling architecture and enjoys structural advantages in navigating the trade-off between reward improvement and constraint satisfaction. To facilitate further research in this area, we reproduce related algorithms in a unified pipeline and incorporate them into SafeRL-Kit, a toolkit that provides off-the-shelf interfaces and evaluation utilities for safety-critical tasks. We then perform a comparative study of the involved algorithms on six benchmarks ranging from robotic control to autonomous driving. The empirical results provide an insight into their applicability and robustness in learning zero-cost-return policies without task-dependent handcrafting. The project page is available at https://sites.google.com/view/saferlkit.

NeurIPS Conference 2023 Conference Paper

Extending the Design Space of Graph Neural Networks by Rethinking Folklore Weisfeiler-Lehman

• Jiarui Feng
• Lecheng Kong
• Hao Liu
• Dacheng Tao
• Fuhai Li
• Muhan Zhang
• Yixin Chen

Message passing neural networks (MPNNs) have emerged as the most popular framework of graph neural networks (GNNs) in recent years. However, their expressive power is limited by the 1-dimensional Weisfeiler-Lehman (1-WL) test. Some works are inspired by $k$-WL/FWL (Folklore WL) and design the corresponding neural versions. Despite the high expressive power, there are serious limitations in this line of research. In particular, (1) $k$-WL/FWL requires at least $O(n^k)$ space complexity, which is impractical for large graphs even when $k=3$; (2) The design space of $k$-WL/FWL is rigid, with the only adjustable hyper-parameter being $k$. To tackle the first limitation, we propose an extension, $(k, t)$-FWL. We theoretically prove that even if we fix the space complexity to $O(n^k)$ (for any $k \geq 2$) in $(k, t)$-FWL, we can construct an expressiveness hierarchy up to solving the graph isomorphism problem. To tackle the second problem, we propose $k$-FWL+, which considers any equivariant set as neighbors instead of all nodes, thereby greatly expanding the design space of $k$-FWL. Combining these two modifications results in a flexible and powerful framework $(k, t)$-FWL+. We demonstrate $(k, t)$-FWL+ can implement most existing models with matching expressiveness. We then introduce an instance of $(k, t)$-FWL+ called Neighborhood$^2$-FWL (N$^2$-FWL), which is practically and theoretically sound. We prove that N$^2$-FWL is no less powerful than 3-WL, and can encode many substructures while only requiring $O(n^2)$ space. Finally, we design its neural version named **N$^2$-GNN** and evaluate its performance on various tasks. N$^2$-GNN achieves record-breaking results on ZINC-Subset (**0. 059**), outperforming previous SOTA results by 10. 6\%. Moreover, N$^2$-GNN achieves new SOTA results on the BREC dataset (**71. 8\%**) among all existing high-expressive GNN methods.

AAAI Conference 2023 Conference Paper

FedABC: Targeting Fair Competition in Personalized Federated Learning

• Dui Wang
• Li Shen
• Yong Luo
• Han Hu
• Kehua Su
• Yonggang Wen
• Dacheng Tao

Federated learning aims to collaboratively train models without accessing their client's local private data. The data may be Non-IID for different clients and thus resulting in poor performance. Recently, personalized federated learning (PFL) has achieved great success in handling Non-IID data by enforcing regularization in local optimization or improving the model aggregation scheme on the server. However, most of the PFL approaches do not take into account the unfair competition issue caused by the imbalanced data distribution and lack of positive samples for some classes in each client. To address this issue, we propose a novel and generic PFL framework termed Federated Averaging via Binary Classification, dubbed FedABC. In particular, we adopt the ``one-vs-all'' training strategy in each client to alleviate the unfair competition between classes by constructing a personalized binary classification problem for each class. This may aggravate the class imbalance challenge and thus a novel personalized binary classification loss that incorporates both the under-sampling and hard sample mining strategies is designed. Extensive experiments are conducted on two popular datasets under different settings, and the results demonstrate that our FedABC can significantly outperform the existing counterparts.

ICLR Conference 2023 Conference Paper

FedSpeed: Larger Local Interval, Less Communication Round, and Higher Generalization Accuracy

• Yan Sun
• Li Shen 0008
• Tiansheng Huang
• Liang Ding 0006
• Dacheng Tao

Federated learning (FL) is an emerging distributed machine learning framework which jointly trains a global model via a large number of local devices with data privacy protections. Its performance suffers from the non-vanishing biases introduced by the local inconsistent optimal and the rugged client-drifts by the local over-fitting. In this paper, we propose a novel and practical method, FedSpeed, to alleviate the negative impacts posed by these problems. Concretely, FedSpeed applies the prox-correction term on the current local updates to efficiently reduce the biases introduced by the prox-term, a necessary regularizer to maintain the strong local consistency. Furthermore, FedSpeed merges the vanilla stochastic gradient with a perturbation computed from an extra gradient ascent step in the neighborhood, thereby alleviating the issue of local over-fitting. Our theoretical analysis indicates that the convergence rate is related to both the communication rounds $T$ and local intervals $K$ with a tighter upper bound $\mathcal{O}(\frac{1}{T})$ if $K=\mathcal{O}(T)$. Moreover, we conduct extensive experiments on the real-world dataset to demonstrate the efficiency of our proposed FedSpeed, which converges significantly faster and achieves the state-of-the-art (SOTA) performance on the general FL experimental settings than several baselines including FedAvg, FedProx, FedCM, FedAdam, SCAFFOLD, FedDyn, FedADMM, etc.

TMLR Journal 2023 Journal Article

Fusion of Global and Local Knowledge for Personalized Federated Learning

• Tiansheng Huang
• Li Shen
• Yan Sun
• Weiwei Lin
• Dacheng Tao

Personalized federated learning, as a variant of federated learning, trains customized models for clients using their heterogeneously distributed data. However, it is still inconclusive about how to design personalized models with better representation of shared global knowledge and personalized pattern. To bridge the gap, we in this paper explore personalized models with low-rank and sparse decomposition. Specifically, we employ proper regularization to extract a low-rank global knowledge representation (GKR), so as to distill global knowledge into a compact representation. Subsequently, we employ a sparse component over the obtained GKR to fuse the personalized pattern into the global knowledge. As a solution, we propose a two-stage proximal-based algorithm named \textbf{Fed}erated learning with mixed \textbf{S}parse and \textbf{L}ow-\textbf{R}ank representation (FedSLR) to efficiently search for the mixed models. Theoretically, under proper assumptions, we show that the GKR trained by FedSLR can at least sub-linearly converge to a stationary point of the regularized problem, and that the sparse component being fused can converge to its stationary point under proper settings. Extensive experiments also demonstrate the superior empirical performance of FedSLR. Moreover, FedSLR reduces the number of parameters, and lowers the down-link communication complexity, which are all desirable for federated learning algorithms. Source code is available in \url{https://github.com/huangtiansheng/fedslr}.

IJCAI Conference 2023 Conference Paper

Graph Pooling for Graph Neural Networks: Progress, Challenges, and Opportunities

• Chuang Liu
• Yibing Zhan
• Jia Wu
• Chang Li
• Bo Du
• Wenbin Hu
• Tongliang Liu
• Dacheng Tao

Graph neural networks have emerged as a leading architecture for many graph-level tasks, such as graph classification and graph generation. As an essential component of the architecture, graph pooling is indispensable for obtaining a holistic graph-level representation of the whole graph. Although a great variety of methods have been proposed in this promising and fast-developing research field, to the best of our knowledge, little effort has been made to systematically summarize these works. To set the stage for the development of future works, in this paper, we attempt to fill this gap by providing a broad review of recent methods for graph pooling. Specifically, 1) we first propose a taxonomy of existing graph pooling methods with a mathematical summary for each category; 2) then, we provide an overview of the libraries related to graph pooling, including the commonly used datasets, model architectures for downstream tasks, and open-source implementations; 3) next, we further outline the applications that incorporate the idea of graph pooling in a variety of domains; 4) finally, we discuss certain critical challenges facing current studies and share our insights on future potential directions for research on the improvement of graph pooling.

ICLR Conference 2023 Conference Paper

HiT-MDP: Learning the SMDP option framework on MDPs with Hidden Temporal Embeddings

• Chang Li
• Dongjin Song
• Dacheng Tao

The standard option framework is developed on the Semi-Markov Decision Process (SMDP) which is unstable to optimize and sample inefficient. To this end, we propose the Hidden Temporal MDP (HiT-MDP) and prove that the option-induced HiT-MDP is homomorphic equivalent to the option-induced SMDP. A novel transformer-based framework is introduced to learn options' embedding vectors (rather than conventional option tuples) on HiT-MDPs. We then derive a stable and sample efficient option discovering method under the maximum-entropy policy gradient framework. Extensive experiments on challenging Mujoco environments demonstrate HiT-MDP's efficiency and effectiveness: under widely used configurations, HiT-MDP achieves competitive, if not better, performance compared to the state-of-the-art baselines on all finite horizon and transfer learning environments. Moreover, HiT-MDP significantly outperforms all baselines on infinite horizon environments while exhibiting smaller variance, faster convergence, and better interpretability. Our work potentially sheds light on the theoretical ground of extending the option framework into a large-scale foundation model.

IJCAI Conference 2023 Conference Paper

Improving Heterogeneous Model Reuse by Density Estimation

• Anke Tang
• Yong Luo
• Han Hu
• Fengxiang He
• Kehua Su
• Bo Du
• Yixin Chen
• Dacheng Tao

This paper studies multiparty learning, aiming to learn a model using the private data of different participants. Model reuse is a promising solution for multiparty learning, assuming that a local model has been trained for each party. Considering the potential sample selection bias among different parties, some heterogeneous model reuse approaches have been developed. However, although pre-trained local classifiers are utilized in these approaches, the characteristics of the local data are not well exploited. This motivates us to estimate the density of local data and design an auxiliary model together with the local classifiers for reuse. To address the scenarios where some local models are not well pre-trained, we further design a multiparty cross-entropy loss for calibration. Upon existing works, we address a challenging problem of heterogeneous model reuse from a decision theory perspective and take advantage of recent advances in density estimation. Experimental results on both synthetic and benchmark data demonstrate the superiority of the proposed method.

AAAI Conference 2023 Conference Paper

Improving Simultaneous Machine Translation with Monolingual Data

• Hexuan Deng
• Liang Ding
• Xuebo Liu
• Meishan Zhang
• Dacheng Tao
• Min Zhang

Simultaneous machine translation (SiMT) is usually done via sequence-level knowledge distillation (Seq-KD) from a full-sentence neural machine translation (NMT) model. However, there is still a significant performance gap between NMT and SiMT. In this work, we propose to leverage monolingual data to improve SiMT, which trains a SiMT student on the combination of bilingual data and external monolingual data distilled by Seq-KD. Preliminary experiments on En-Zh and En-Ja news domain corpora demonstrate that monolingual data can significantly improve translation quality (e.g., +3.15 BLEU on En-Zh). Inspired by the behavior of human simultaneous interpreters, we propose a novel monolingual sampling strategy for SiMT, considering both chunk length and monotonicity. Experimental results show that our sampling strategy consistently outperforms the random sampling strategy (and other conventional typical NMT monolingual sampling strategies) by avoiding the key problem of SiMT -- hallucination, and has better scalability. We achieve +0.72 BLEU improvements on average against random sampling on En-Zh and En-Ja. Data and codes can be found at https://github.com/hexuandeng/Mono4SiMT.

ICML Conference 2023 Conference Paper

Improving the Model Consistency of Decentralized Federated Learning

• Yifan Shi
• Li Shen 0008
• Kang Wei 0004
• Yan Sun
• Bo Yuan 0003
• Xueqian Wang 0001
• Dacheng Tao

To mitigate the privacy leakages and communication burdens of Federated Learning (FL), decentralized FL (DFL) discards the central server and each client only communicates with its neighbors in a decentralized communication network. However, existing DFL suffers from high inconsistency among local clients, which results in severe distribution shift and inferior performance compared with centralized FL (CFL), especially on heterogeneous data or sparse communication topologies. To alleviate this issue, we propose two DFL algorithms named DFedSAM and DFedSAM-MGS to improve the performance of DFL. Specifically, DFedSAM leverages gradient perturbation to generate local flat models via Sharpness Aware Minimization (SAM), which searches for models with uniformly low loss values. DFedSAM-MGS further boosts DFedSAM by adopting Multiple Gossip Steps (MGS) for better model consistency, which accelerates the aggregation of local flat models and better balances communication complexity and generalization. Theoretically, we present improved convergence rates $\small \mathcal{O}\big(\frac{1}{\sqrt{KT}}+\frac{1}{T}+\frac{1}{K^{1/2}T^{3/2}(1-\lambda)^2}\big)$ and $\small \mathcal{O}\big(\frac{1}{\sqrt{KT}}+\frac{1}{T}+\frac{\lambda^Q+1}{K^{1/2}T^{3/2}(1-\lambda^Q)^2}\big)$ in non-convex setting for DFedSAM and DFedSAM-MGS, respectively, where $1-\lambda$ is the spectral gap of gossip matrix and $Q$ is the number of MGS. Empirically, our methods can achieve competitive performance compared with CFL methods and outperform existing DFL methods.

NeurIPS Conference 2023 Conference Paper

Learning Better with Less: Effective Augmentation for Sample-Efficient Visual Reinforcement Learning

• Guozheng Ma
• Linrui Zhang
• Haoyu Wang
• Lu Li
• Zilin Wang
• Zhen Wang
• Li Shen
• Xueqian Wang

Data augmentation (DA) is a crucial technique for enhancing the sample efficiency of visual reinforcement learning (RL) algorithms. Notably, employing simple observation transformations alone can yield outstanding performance without extra auxiliary representation tasks or pre-trained encoders. However, it remains unclear which attributes of DA account for its effectiveness in achieving sample-efficient visual RL. To investigate this issue and further explore the potential of DA, this work conducts comprehensive experiments to assess the impact of DA's attributes on its efficacy and provides the following insights and improvements: (1) For individual DA operations, we reveal that both ample spatial diversity and slight hardness are indispensable. Building on this finding, we introduce Random PadResize (Rand PR), a new DA operation that offers abundant spatial diversity with minimal hardness. (2) For multi-type DA fusion schemes, the increased DA hardness and unstable data distribution result in the current fusion schemes being unable to achieve higher sample efficiency than their corresponding individual operations. Taking the non-stationary nature of RL into account, we propose a RL-tailored multi-type DA fusion scheme called Cycling Augmentation (CycAug), which performs periodic cycles of different DA operations to increase type diversity while maintaining data distribution consistency. Extensive evaluations on the DeepMind Control suite and CARLA driving simulator demonstrate that our methods achieve superior sample efficiency compared with the prior state-of-the-art methods.

AAAI Conference 2023 Conference Paper

Learning to Learn Better for Video Object Segmentation

• Meng Lan
• Jing Zhang
• Lefei Zhang
• Dacheng Tao

Recently, the joint learning framework (JOINT) integrates matching based transductive reasoning and online inductive learning to achieve accurate and robust semi-supervised video object segmentation (SVOS). However, using the mask embedding as the label to guide the generation of target features in the two branches may result in inadequate target representation and degrade the performance. Besides, how to reasonably fuse the target features in the two different branches rather than simply adding them together to avoid the adverse effect of one dominant branch has not been investigated. In this paper, we propose a novel framework that emphasizes Learning to Learn Better (LLB) target features for SVOS, termed LLB, where we design the discriminative label generation module (DLGM) and the adaptive fusion module to address these issues. Technically, the DLGM takes the background-filtered frame instead of the target mask as input and adopts a lightweight encoder to generate the target features, which serves as the label of the online few-shot learner and the value of the decoder in the transformer to guide the two branches to learn more discriminative target representation. The adaptive fusion module maintains a learnable gate for each branch, which reweighs the element-wise feature representation and allows an adaptive amount of target information in each branch flowing to the fused target feature, thus preventing one branch from being dominant and making the target feature more robust to distractor. Extensive experiments on public benchmarks show that our proposed LLB method achieves state-of-the-art performance.

ICML Conference 2023 Conference Paper

Learning to Learn from APIs: Black-Box Data-Free Meta-Learning

• Zixuan Hu
• Li Shen 0008
• Zhenyi Wang 0001
• Baoyuan Wu
• Chun Yuan 0003
• Dacheng Tao

Data-free meta-learning (DFML) aims to enable efficient learning of new tasks by meta-learning from a collection of pre-trained models without access to the training data. Existing DFML work can only meta-learn from (i) white-box and (ii) small-scale pre-trained models (iii) with the same architecture, neglecting the more practical setting where the users only have inference access to the APIs with arbitrary model architectures and model scale inside. To solve this issue, we propose a Bi-level Data-free Meta Knowledge Distillation (BiDf-MKD) framework to transfer more general meta knowledge from a collection of black-box APIs to one single meta model. Specifically, by just querying APIs, we inverse each API to recover its training data via a zero-order gradient estimator and then perform meta-learning via a novel bi-level meta knowledge distillation structure, in which we design a boundary query set recovery technique to recover a more informative query set near the decision boundary. In addition, to encourage better generalization within the setting of limited API budgets, we propose task memory replay to diversify the underlying task distribution by covering more interpolated tasks. Extensive experiments in various real-world scenarios show the superior performance of our BiDf-MKD framework.

NeurIPS Conference 2023 Conference Paper

MAG-GNN: Reinforcement Learning Boosted Graph Neural Network

• Lecheng Kong
• Jiarui Feng
• Hao Liu
• Dacheng Tao
• Yixin Chen
• Muhan Zhang

While Graph Neural Networks (GNNs) recently became powerful tools in graph learning tasks, considerable efforts have been spent on improving GNNs' structural encoding ability. A particular line of work proposed subgraph GNNs that use subgraph information to improve GNNs' expressivity and achieved great success. However, such effectivity sacrifices the efficiency of GNNs by enumerating all possible subgraphs. In this paper, we analyze the necessity of complete subgraph enumeration and show that a model can achieve a comparable level of expressivity by considering a small subset of the subgraphs. We then formulate the identification of the optimal subset as a combinatorial optimization problem and propose Magnetic Graph Neural Network (MAG-GNN), a reinforcement learning (RL) boosted GNN, to solve the problem. Starting with a candidate subgraph set, MAG-GNN employs an RL agent to iteratively update the subgraphs to locate the most expressive set for prediction. This reduces the exponential complexity of subgraph enumeration to the constant complexity of a subgraph search algorithm while keeping good expressivity. We conduct extensive experiments on many datasets, showing that MAG-GNN achieves competitive performance to state-of-the-art methods and even outperforms many subgraph GNNs. We also demonstrate that MAG-GNN effectively reduces the running time of subgraph GNNs.

ICLR Conference 2023 Conference Paper

Neural Collapse Inspired Feature-Classifier Alignment for Few-Shot Class-Incremental Learning

• Yibo Yang
• Haobo Yuan
• Xiangtai Li
• Zhouchen Lin
• Philip H. S. Torr
• Dacheng Tao

Few-shot class-incremental learning (FSCIL) has been a challenging problem as only a few training samples are accessible for each novel class in the new sessions. Finetuning the backbone or adjusting the classifier prototypes trained in the prior sessions would inevitably cause a misalignment between the feature and classifier of old classes, which explains the well-known catastrophic forgetting problem. In this paper, we deal with this misalignment dilemma in FSCIL inspired by the recently discovered phenomenon named neural collapse, which reveals that the last-layer features of the same class will collapse into a vertex, and the vertices of all classes are aligned with the classifier prototypes, which are formed as a simplex equiangular tight frame (ETF). It corresponds to an optimal geometric structure for classification due to the maximized Fisher Discriminant Ratio. We propose a neural collapse inspired framework for FSCIL. A group of classifier prototypes are pre-assigned as a simplex ETF for the whole label space, including the base session and all the incremental sessions. During training, the classifier prototypes are not learnable, and we adopt a novel loss function that drives the features into their corresponding prototypes. Theoretical analysis shows that our method holds the neural collapse optimality and does not break the feature-classifier alignment in an incremental fashion. Experiments on the miniImageNet, CUB-200, and CIFAR-100 datasets demonstrate that our proposed framework outperforms the state-of-the-art performances. Code address: https://github.com/NeuralCollapseApplications/FSCIL

ICRA Conference 2023 Conference Paper

Off-policy Imitation Learning from Visual Inputs

• Zhihao Cheng
• Li Shen 0008
• Dacheng Tao

Recently, various successful applications utilizing expert states in imitation learning (IL) have been witnessed. However, IL from visual inputs (ILfVI), which has a greater promise to be widely applied by using online visual resources, suffers from low data-efficiency and poor performance resulted from on-policy learning and high-dimensional visual inputs. We propose OPIfVI (Off-Policy Imitation from Visual Inputs), which is composed of an off-policy learning manner, data augmentation, and encoder techniques, to tackle the mentioned challenges, respectively. More specifically, to improve data-efficiency, OPIfVI conducts IL in an off-policy manner, with which sampled data used multiple times. In addition, we enhance the stability of OPIfVI with spectral normalization to mitigate the side effect of off-policy training. The core factor, contributing to the poor performance of ILfVI, that we think is agents could not extract meaningful features from visual inputs. Hence, OPIfVI employs data augmentation from computer vision to help train encoders to better extract features from visual inputs. Besides, a specific structure of gradient backpropagation for the encoder is designed to stabilize the encoder training. At last, we demonstrate that OPIfVI can achieve expert-level performance and outperform existing baselines via extensive experiments using DeepMind Control Suite.

AAAI Conference 2023 Conference Paper

Offline Quantum Reinforcement Learning in a Conservative Manner

• Zhihao Cheng
• Kaining Zhang
• Li Shen
• Dacheng Tao

Recently, to reap the quantum advantage, empowering reinforcement learning (RL) with quantum computing has attracted much attention, which is dubbed as quantum RL (QRL). However, current QRL algorithms employ an online learning scheme, i.e., the policy that is run on a quantum computer needs to interact with the environment to collect experiences, which could be expensive and dangerous for practical applications. In this paper, we aim to solve this problem in an offline learning manner. To be more specific, we develop the first offline quantum RL (offline QRL) algorithm named CQ2L (Conservative Quantum Q-learning), which learns from offline samples and does not require any interaction with the environment. CQ2L utilizes variational quantum circuits (VQCs), which are improved with data re-uploading and scaling parameters, to represent Q-value functions of agents. To suppress the overestimation of Q-values resulting from offline data, we first employ a double Q-learning framework to reduce the overestimation bias; then a penalty term that encourages generating conservative Q-values is designed. We conduct abundant experiments to demonstrate that the proposed method CQ2L can successfully solve offline QRL tasks that the online counterpart could not.

EAAI Journal 2023 Journal Article

RSAL-iMFS: A framework of randomized stacking with active learning for incremental multi-fidelity surrogate modeling

• Zongqi Liu
• Xueguan Song
• Chao Zhang
• Yunsheng Ma
• Dacheng Tao

Multi-fidelity surrogate (MFS) modeling incorporates a large number of low-fidelity (LF) samples with a small size of high-fidelity (HF) samples to obtain accurate HF approximations of unknown inputs. Because they save costs when labeling massive HF samples, MFS modeling techniques have been widely used in many engineering problems such as rapid simulation and mechanism optimization. In this paper, we propose a framework of randomized stacking with active learning for incremental MFS modeling (RSAL-iMFS). The framework contains two parts: randomized stacking for incremental MFS modeling (RS-iMFS) and query-by-committee (QBC)-based active learning for MFS modeling (QBC-AL-MFS). In particular, we first randomly project the inputs of LF samples into different spaces spanned by random projection matrices to form new LF samples and build a series of LF regressors based on the transformed LF samples to capture the LF features from different views. Then, these base LF regressors are stacked to form the inputs of the subsequent incremental Gaussian process regression (iGPR) model for approximating the HF responses. To achieve improved modeling performance, we also adopt the QBC-based active learning (QBC-AL) method to recommend unlabeled inputs from the input pool of candidate HF samples according to their distances, which are weighted by using the divergence of the committee voting results. We then label these recommended inputs to incrementally update the current iGPR model. Numerical experiments validate the proposed framework in incremental MFS modeling tasks and show that (1) RSAL-iMFS outperforms the state-of-the-art models; (2) Gaussian projection matrices (GPMs) can be reasonable choices in most cases; and (3) RSAL-iMFS’s modeling performance has low sensitivity to the choice of the number of base regressors.

NeurIPS Conference 2023 Conference Paper

SAMRS: Scaling-up Remote Sensing Segmentation Dataset with Segment Anything Model

• Di Wang
• Jing Zhang
• Bo Du
• Minqiang Xu
• Lin Liu
• Dacheng Tao
• Liangpei Zhang

The success of the Segment Anything Model (SAM) demonstrates the significance of data-centric machine learning. However, due to the difficulties and high costs associated with annotating Remote Sensing (RS) images, a large amount of valuable RS data remains unlabeled, particularly at the pixel level. In this study, we leverage SAM and existing RS object detection datasets to develop an efficient pipeline for generating a large-scale RS segmentation dataset, dubbed SAMRS. SAMRS totally possesses 105, 090 images and 1, 668, 241 instances, surpassing existing high-resolution RS segmentation datasets in size by several orders of magnitude. It provides object category, location, and instance information that can be used for semantic segmentation, instance segmentation, and object detection, either individually or in combination. We also provide a comprehensive analysis of SAMRS from various aspects. Moreover, preliminary experiments highlight the importance of conducting segmentation pre-training with SAMRS to address task discrepancies and alleviate the limitations posed by limited training data during fine-tuning. The code and dataset will be available at https: //github. com/ViTAE-Transformer/SAMRS

ICLR Conference 2023 Conference Paper

Sharper Bounds for Uniformly Stable Algorithms with Stationary Mixing Process

• Shi Fu
• Yunwen Lei
• Qiong Cao
• Xinmei Tian 0001
• Dacheng Tao

Generalization analysis of learning algorithms often builds on a critical assumption that training examples are independently and identically distributed, which is often violated in practical problems such as time series prediction. In this paper, we use algorithmic stability to study the generalization performance of learning algorithms with $\psi$-mixing data, where the dependency between observations weakens over time. We show uniformly stable algorithms guarantee high-probability generalization bounds of the order $O(1/\sqrt{n})$ (within a logarithmic factor), where $n$ is the sample size. We apply our general result to specific algorithms including regularization schemes, stochastic gradient descent and localized iterative regularization, and develop excess population risk bounds for learning with $\psi$-mixing data. Our analysis builds on a novel moment bound for weakly-dependent random variables on a $\varphi$-mixing sequence and a novel error decomposition of generalization error.

AIJ Journal 2023 Journal Article

Spectral complexity-scaled generalisation bound of complex-valued neural networks

• Haowen Chen
• Fengxiang He
• Shiye Lei
• Dacheng Tao

Complex-valued neural networks (CVNNs) have been widely applied in various fields, primarily in signal processing and image recognition. Few studies have focused on the generalisation of CVNNs, although it is vital to ensure the performance of CVNNs on unseen data. This study is the first to prove a generalisation bound for complex-valued neural networks. The bounds increase as the spectral complexity increases, with the dominant factor being the product of the spectral norms of the weight matrices. Furthermore, this work provides a generalisation bound for CVNNs trained on sequential data, which is also affected by the spectral complexity. Theoretically, these bounds are derived using the Maurey Sparsification Lemma and Dudley entropy integral. We conducted empirical experiments on various datasets including MNIST, ashionMNIST, CIFAR-10, CIFAR-100, Tiny ImageNet, and IMDB by training complex-valued convolutional neural networks. The Spearman rank-order correlation coefficient and the corresponding p-values on these datasets provide strong proof of the statistically significant correlation between the spectral complexity of a network and its generalisation ability, as measured by the spectral norm product of the weight matrices. The code is available at https: //github. com/LeavesLei/cvnn_generalization.

NeurIPS Conference 2023 Conference Paper

Stability and Generalization of the Decentralized Stochastic Gradient Descent Ascent Algorithm

• Miaoxi Zhu
• Li Shen
• Bo Du
• Dacheng Tao

The growing size of available data has attracted increasing interest in solving minimax problems in a decentralized manner for various machine learning tasks. Previous theoretical research has primarily focused on the convergence rate and communication complexity of decentralized minimax algorithms, with little attention given to their generalization. In this paper, we investigate the primal-dual generalization bound of the decentralized stochastic gradient descent ascent (D-SGDA) algorithm using the approach of algorithmic stability under both convex-concave and nonconvex-nonconcave settings. Our theory refines the algorithmic stability in a decentralized manner and demonstrates that the decentralized structure does not destroy the stability and generalization of D-SGDA, implying that it can generalize as well as the vanilla SGDA in certain situations. Our results analyze the impact of different topologies on the generalization bound of the D-SGDA algorithm beyond trivial factors such as sample sizes, learning rates, and iterations. We also evaluate the optimization error and balance it with the generalization gap to obtain the optimal population risk of D-SGDA in the convex-concave setting. Additionally, we perform several numerical experiments which validate our theoretical findings.

ICML Conference 2023 Conference Paper

Structured Cooperative Learning with Graphical Model Priors

• Shuangtong Li
• Tianyi Zhou 0001
• Xinmei Tian 0001
• Dacheng Tao

We study how to train personalized models for different tasks on decentralized devices with limited local data. We propose "Structured Cooperative Learning (SCooL)", in which a cooperation graph across devices is generated by a graphical model prior to automatically coordinate mutual learning between devices. By choosing graphical models enforcing different structures, we can derive a rich class of existing and novel decentralized learning algorithms via variational inference. In particular, we show three instantiations of SCooL that adopt Dirac distribution, stochastic block model (SBM), and attention as the prior generating cooperation graphs. These EM-type algorithms alternate between updating the cooperation graph and cooperative learning of local models. They can automatically capture the cross-task correlations among devices by only monitoring their model updating in order to optimize the cooperation graph. We evaluate SCooL and compare it with existing decentralized learning methods on an extensive set of benchmarks, on which SCooL always achieves the highest accuracy of personalized models and significantly outperforms other baselines on communication efficiency. Our code is available at https: //github. com/ShuangtongLi/SCooL.

ICLR Conference 2023 Conference Paper

Symmetric Pruning in Quantum Neural Networks

• Xinbiao Wang
• Junyu Liu
• Tongliang Liu
• Yong Luo 0002
• Yuxuan Du
• Dacheng Tao

Many fundamental properties of a quantum system are captured by its Hamiltonian and ground state. Despite the significance, ground states preparation (GSP) is classically intractable for large-scale Hamiltonians. Quantum neural networks (QNNs), which exert the power of modern quantum machines, have emerged as a leading protocol to conquer this issue. As such, the performance enhancement of QNNs becomes the core in GSP. Empirical evidence showed that QNNs with handcraft symmetric ans\"atze generally experience better trainability than those with asymmetric ans\"atze, while theoretical explanations remain vague. To fill this knowledge gap, here we propose the effective quantum neural tangent kernel (EQNTK) and connect this concept with over-parameterization theory to quantify the convergence of QNNs towards the global optima. We uncover that the advance of symmetric ans\"atze attributes to their large EQNTK value with low effective dimension, which requests few parameters and quantum circuit depth to reach the over-parameterization regime permitting a benign loss landscape and fast convergence. Guided by EQNTK, we further devise a symmetric pruning (SP) scheme to automatically tailor a symmetric ansatz from an over-parameterized and asymmetric one to greatly improve the performance of QNNs when the explicit symmetry information of Hamiltonian is unavailable. Extensive numerical simulations are conducted to validate the analytical results of EQNTK and the effectiveness of SP.

ICML Conference 2023 Conference Paper

Tilted Sparse Additive Models

• Yingjie Wang 0007
• Hong Chen 0004
• Weifeng Liu 0001
• Fengxiang He
• Tieliang Gong
• Youcheng Fu
• Dacheng Tao

Additive models have been burgeoning in data analysis due to their flexible representation and desirable interpretability. However, most existing approaches are constructed under empirical risk minimization (ERM), and thus perform poorly in situations where average performance is not a suitable criterion for the problems of interest, e. g. , data with complex non-Gaussian noise, imbalanced labels or both of them. In this paper, a novel class of sparse additive models is proposed under tilted empirical risk minimization (TERM), which addresses the deficiencies in ERM by imposing tilted impact on individual losses, and is flexibly capable of achieving a variety of learning objectives, e. g. , variable selection, robust estimation, imbalanced classification and multiobjective learning. On the theoretical side, a learning theory analysis which is centered around the generalization bound and function approximation error bound (under some specific data distributions) is conducted rigorously. On the practical side, an accelerated optimization algorithm is designed by integrating Prox-SVRG and random Fourier acceleration technique. The empirical assessments verify the competitive performance of our approach on both synthetic and real data.

ICLR Conference 2023 Conference Paper

Towards One-shot Neural Combinatorial Solvers: Theoretical and Empirical Notes on the Cardinality-Constrained Case

• Runzhong Wang
• Li Shen 0008
• Yiting Chen 0003
• Xiaokang Yang 0001
• Dacheng Tao
• Junchi Yan

One-shot non-autoregressive neural networks, different from RL-based ones, have been actively adopted for solving combinatorial optimization (CO) problems, which can be trained by the objective score in a self-supervised manner. Such methods have shown their superiority in efficiency (e.g. by parallelization) and potential for tackling predictive CO problems for decision-making under uncertainty. While the discrete constraints often become a bottleneck for gradient-based neural solvers, as currently handled in three typical ways: 1) adding a soft penalty in the objective, where a bounded violation of the constraints cannot be guaranteed, being critical to many constraint-sensitive scenarios; 2) perturbing the input to generate an approximate gradient in a black-box manner, though the constraints are exactly obeyed while the approximate gradients can hurt the performance on the objective score; 3) a compromise by developing soft algorithms whereby the output of neural networks obeys a relaxed constraint, and there can still occur an arbitrary degree of constraint-violation. Towards the ultimate goal of establishing a general framework for neural CO solver with the ability to control an arbitrary-small degree of constraint violation, in this paper, we focus on a more achievable and common setting: the cardinality constraints, which in fact can be readily encoded by a differentiable optimal transport (OT) layer. Based on this observation, we propose OT-based cardinality constraint encoding for end-to-end CO problem learning with two variants: Sinkhorn and Gumbel-Sinkhorn, whereby their violation of the constraints can be exactly characterized and bounded by our theoretical results. On synthetic and real-world CO problem instances, our methods surpass the state-of-the-art CO network and are comparable to (if not superior to) the commercial solver Gurobi. In particular, we further showcase a case study of applying our approach to the predictive portfolio optimization task on real-world asset price data, improving the Sharpe ratio from 1.1 to 2.0 of a strong LSTM+Gurobi baseline under the classic predict-then-optimize paradigm.

NeurIPS Conference 2023 Conference Paper

Understanding How Consistency Works in Federated Learning via Stage-wise Relaxed Initialization

• Yan Sun
• Li Shen
• Dacheng Tao

Federated learning (FL) is a distributed paradigm that coordinates massive local clients to collaboratively train a global model via stage-wise local training processes on the heterogeneous dataset. Previous works have implicitly studied that FL suffers from the "client-drift" problem, which is caused by the inconsistent optimum across local clients. However, till now it still lacks solid theoretical analysis to explain the impact of this local inconsistency. To alleviate the negative impact of the "client drift" and explore its substance in FL, in this paper, we first design an efficient FL algorithm FedInit, which allows employing the personalized relaxed initialization state at the beginning of each local training stage. Specifically, FedInit initializes the local state by moving away from the current global state towards the reverse direction of the latest local state. This relaxed initialization helps to revise the local divergence and enhance the local consistency level. Moreover, to further understand how inconsistency disrupts performance in FL, we introduce the excess risk analysis and study the divergence term to investigate the test error of the proposed FedInit method. Our studies show that on the non-convex objectives, optimization error is not sensitive to this local inconsistency, while it mainly affects the generalization error bound in FedInit. Extensive experiments are conducted to validate this conclusion. Our proposed FedInit could achieve state-of-the-art (SOTA) results compared to several advanced benchmarks without any additional costs. Meanwhile, stage-wise relaxed initialization could also be incorporated into the current advanced algorithms to achieve higher performance in the FL paradigm.

ICLR Conference 2023 Conference Paper

Unified Discrete Diffusion for Simultaneous Vision-Language Generation

• Minghui Hu 0001
• Chuanxia Zheng
• Zuopeng Yang
• Tat-Jen Cham
• Heliang Zheng
• Chaoyue Wang
• Dacheng Tao
• Ponnuthurai N. Suganthan

The recently developed discrete diffusion model performs extraordinarily well in generation tasks, especially in the text-to-image task, showing great potential for modeling multimodal signals. In this paper, we leverage these properties and present a unified multimodal generation model, which can perform text-based, image-based, and even vision-language simultaneous generation using a single model. Specifically, we unify the discrete diffusion process for multimodal signals by proposing a unified Markov transition matrix and a unified objective. Moreover, we design a multimodal mutual attention module to highlight the inter-modal linkages, which is vital for multimodal generation. Extensive experiments indicate that our proposed method can perform comparably to the state-of-the-art solutions in various generation tasks.

NeurIPS Conference 2023 Conference Paper

VanillaNet: the Power of Minimalism in Deep Learning

• Hanting Chen
• Yunhe Wang
• Jianyuan Guo
• Dacheng Tao

At the heart of foundation models is the philosophy of "more is different", exemplified by the astonishing success in computer vision and natural language processing. However, the challenges of optimization and inherent complexity of transformer models call for a paradigm shift towards simplicity. In this study, we introduce VanillaNet, a neural network architecture that embraces elegance in design. By avoiding high depth, shortcuts, and intricate operations like self-attention, VanillaNet is refreshingly concise yet remarkably powerful. Each layer is carefully crafted to be compact and straightforward, with nonlinear activation functions pruned after training to restore the original architecture. VanillaNet overcomes the challenges of inherent complexity, making it ideal for resource-constrained environments. Its easy-to-understand and highly simplified architecture opens new possibilities for efficient deployment. Extensive experimentation demonstrates that VanillaNet delivers performance on par with renowned deep neural networks and vision transformers, showcasing the power of minimalism in deep learning. This visionary journey of VanillaNet has significant potential to redefine the landscape and challenge the status quo of foundation model, setting a new path for elegant and effective model design. Pre-trained models and codes are available at https: //github. com/huawei-noah/VanillaNet and https: //gitee. com/mindspore/models/tree/master/research/cv/vanillanet

ICLR Conference 2022 Conference Paper

A Relational Intervention Approach for Unsupervised Dynamics Generalization in Model-Based Reinforcement Learning

• Jiaxian Guo
• Mingming Gong
• Dacheng Tao

The generalization of model-based reinforcement learning (MBRL) methods to environments with unseen transition dynamics is an important yet challenging problem. Existing methods try to extract environment-specified information $Z$ from past transition segments to make the dynamics prediction model generalizable to different dynamics. However, because environments are not labelled, the extracted information inevitably contains redundant information unrelated to the dynamics in transition segments and thus fails to maintain a crucial property of $Z$: $Z$ should be similar in the same environment and dissimilar in different ones. As a result, the learned dynamics prediction function will deviate from the true one, which undermines the generalization ability. To tackle this problem, we introduce an interventional prediction module to estimate the probability of two estimated $\hat{z}_i, \hat{z}_j$ belonging to the same environment. Furthermore, by utilizing the $Z$'s invariance within a single environment, a relational head is proposed to enforce the similarity between $\hat{{Z}}$ from the same environment. As a result, the redundant information will be reduced in $\hat{Z}$. We empirically show that $\hat{{Z}}$ estimated by our method enjoy less redundant information than previous methods, and such $\hat{{Z}}$ can significantly reduce dynamics prediction errors and improve the performance of model-based RL methods on zero-shot new environments with unseen dynamics. The codes of this method are available at \url{https://github.com/CR-Gjx/RIA}.

IJCAI Conference 2022 Conference Paper

A Survey on Gradient Inversion: Attacks, Defenses and Future Directions

• Rui Zhang
• Song Guo
• Junxiao Wang
• Xin Xie
• Dacheng Tao

Recent studies have shown that the training samples can be recovered from gradients, which are called Gradient Inversion (GradInv) attacks. However, there remains a lack of extensive surveys covering recent advances and thorough analysis of this issue. In this paper, we present a comprehensive survey on GradInv, aiming to summarize the cutting-edge research and broaden the horizons for different domains. Firstly, we propose a taxonomy of GradInv attacks by characterizing existing attacks into two paradigms: iteration- and recursion-based attacks. In particular, we dig out some critical ingredients from the iteration-based attacks, including data initialization, model training and gradient matching. Second, we summarize emerging defense strategies against GradInv attacks. We find these approaches focus on three perspectives covering data obscuration, model improvement and gradient protection. Finally, we discuss some promising directions and open problems for further research.

NeurIPS Conference 2022 Conference Paper

Adversarial Auto-Augment with Label Preservation: A Representation Learning Principle Guided Approach

• Kaiwen Yang
• Yanchao Sun
• Jiahao Su
• Fengxiang He
• Xinmei Tian
• Furong Huang
• Tianyi Zhou
• Dacheng Tao

Data augmentation is a critical contributing factor to the success of deep learning but heavily relies on prior domain knowledge which is not always available. Recent works on automatic data augmentation learn a policy to form a sequence of augmentation operations, which are still pre-defined and restricted to limited options. In this paper, we show that a prior-free autonomous data augmentation's objective can be derived from a representation learning principle that aims to preserve the minimum sufficient information of the labels. Given an example, the objective aims at creating a distant ``hard positive example'' as the augmentation, while still preserving the original label. We then propose a practical surrogate to the objective that can be optimized efficiently and integrated seamlessly into existing methods for a broad class of machine learning tasks, e. g. , supervised, semi-supervised, and noisy-label learning. Unlike previous works, our method does not require training an extra generative model but instead leverages the intermediate layer representations of the end-task model for generating data augmentations. In experiments, we show that our method consistently brings non-trivial improvements to the three aforementioned learning tasks from both efficiency and final performance, either or not combined with pre-defined augmentations, e. g. , on medical images when domain knowledge is unavailable and the existing augmentation techniques perform poorly. Code will be released publicly.

NeurIPS Conference 2022 Conference Paper

APT-36K: A Large-scale Benchmark for Animal Pose Estimation and Tracking

• Yuxiang Yang
• Junjie Yang
• Yufei Xu
• Jing Zhang
• Long Lan
• Dacheng Tao

Animal pose estimation and tracking (APT) is a fundamental task for detecting and tracking animal keypoints from a sequence of video frames. Previous animal-related datasets focus either on animal tracking or single-frame animal pose estimation, and never on both aspects. The lack of APT datasets hinders the development and evaluation of video-based animal pose estimation and tracking methods, limiting the applications in real world, e. g. , understanding animal behavior in wildlife conservation. To fill this gap, we make the first step and propose APT-36K, i. e. , the first large-scale benchmark for animal pose estimation and tracking. Specifically, APT-36K consists of 2, 400 video clips collected and filtered from 30 animal species with 15 frames for each video, resulting in 36, 000 frames in total. After manual annotation and careful double-check, high-quality keypoint and tracking annotations are provided for all the animal instances. Based on APT-36K, we benchmark several representative models on the following three tracks: (1) supervised animal pose estimation on a single frame under intra- and inter-domain transfer learning settings, (2) inter-species domain generalization test for unseen animals, and (3) animal pose estimation with animal tracking. Based on the experimental results, we gain some empirical insights and show that APT-36K provides a useful animal pose estimation and tracking benchmark, offering new challenges and opportunities for future research. The code and dataset will be made publicly available at https: //github. com/pandorgan/APT-36K.

NeurIPS Conference 2022 Conference Paper

Benefits of Permutation-Equivariance in Auction Mechanisms

• Tian Qin
• Fengxiang He
• Dingfeng Shi
• Wenbing Huang
• Dacheng Tao

Designing an incentive-compatible auction mechanism that maximizes the auctioneer's revenue while minimizes the bidders’ ex-post regret is an important yet intricate problem in economics. Remarkable progress has been achieved through learning the optimal auction mechanism by neural networks. In this paper, we consider the popular additive valuation and symmetric valuation setting; i. e. , the valuation for a set of items is defined as the sum of all items’ valuations in the set, and the valuation distribution is invariant when the bidders and/or the items are permutated. We prove that permutation-equivariant neural networks have significant advantages: the permutation-equivariance decreases the expected ex-post regret, improves the model generalizability, while maintains the expected revenue invariant. This implies that the permutation-equivariance helps approach the theoretically optimal dominant strategy incentive compatible condition, and reduces the required sample complexity for desired generalization. Extensive experiments fully support our theory. To our best knowledge, this is the first work towards understanding the benefits of permutation-equivariance in auction mechanisms.

NeurIPS Conference 2022 Conference Paper

CGLB: Benchmark Tasks for Continual Graph Learning

• Xikun Zhang
• Dongjin Song
• Dacheng Tao

Continual learning on graph data, which aims to accommodate new tasks over newly emerged graph data while maintaining the model performance over existing tasks, is attracting increasing attention from the community. Unlike continual learning on Euclidean data ($\textit{e. g. }$, images, texts, etc. ) that has established benchmarks and unified experimental settings, benchmark tasks are rare for Continual Graph Learning (CGL). Moreover, due to the variety of graph data and its complex topological structures, existing works adopt different protocols to configure datasets and experimental settings. This creates a great obstacle to compare different techniques and thus hinders the development of CGL. To this end, we systematically study the task configurations in different application scenarios and develop a comprehensive Continual Graph Learning Benchmark (CGLB) curated from different public datasets. Specifically, CGLB contains both node-level and graph-level continual graph learning tasks under task-incremental (currently widely adopted) and class-incremental (more practical, challenging, yet underexplored) settings, as well as a toolkit for training, evaluating, and visualizing different CGL methods. Within CGLB, we also systematically explain the difference among these task configurations by comparing them to classical continual learning settings. Finally, we comprehensively compare state-of-the-art baselines on CGLB to investigate their effectiveness. Given CGLB and the developed toolkit, the barrier to exploring CGL has been greatly lowered and researchers can focus more on the model development without worrying about tedious work on pre-processing of datasets or encountering unseen pitfalls. The benchmark and the toolkit are available through https: //github. com/QueuQ/CGLB.

AAAI Conference 2022 Conference Paper

Continual Learning through Retrieval and Imagination

• Zhen Wang
• Liu Liu
• Yiqun Duan
• Dacheng Tao

Continual learning is an intellectual ability of artificial agents to learn new streaming labels from sequential data. The main impediment to continual learning is catastrophic forgetting, a severe performance degradation on previously learned tasks. Although simply replaying all previous data or continuously adding the model parameters could alleviate the issue, it is impractical in real-world applications due to the limited available resources. Inspired by the mechanism of the human brain to deepen its past impression, we propose a novel framework, Deep Retrieval and Imagination (DRI), which consists of two components: 1) an embedding network that constructs a unified embedding space without adding model parameters on the arrival of new tasks; and 2) a generative model to produce additional (imaginary) data based on the limited memory. By retrieving the past experiences and corresponding imaginary data, DRI distills knowledge and rebalances the embedding space to further mitigate forgetting. Theoretical analysis demonstrates that DRI can reduce the loss approximation error and improve the robustness through retrieval and imagination, bringing better generalizability to the network. Extensive experiments show that DRI performs significantly better than the existing state-of-the-art continual learning methods and effectively alleviates catastrophic forgetting.

ICML Conference 2022 Conference Paper

DisPFL: Towards Communication-Efficient Personalized Federated Learning via Decentralized Sparse Training

• Rong Dai
• Li Shen 0008
• Fengxiang He
• Xinmei Tian 0001
• Dacheng Tao

Personalized federated learning is proposed to handle the data heterogeneity problem amongst clients by learning dedicated tailored local models for each user. However, existing works are often built in a centralized way, leading to high communication pressure and high vulnerability when a failure or an attack on the central server occurs. In this work, we propose a novel personalized federated learning framework in a decentralized (peer-to-peer) communication protocol named DisPFL, which employs personalized sparse masks to customize sparse local models on the edge. To further save the communication and computation cost, we propose a decentralized sparse training technique, which means that each local model in DisPFL only maintains a fixed number of active parameters throughout the whole local training and peer-to-peer communication process. Comprehensive experiments demonstrate that DisPFL significantly saves the communication bottleneck for the busiest node among all clients and, at the same time, achieves higher model accuracy with less computation cost and communication rounds. Furthermore, we demonstrate that our method can easily adapt to heterogeneous local clients with varying computation complexities and achieves better personalized performances.

NeurIPS Conference 2022 Conference Paper

Escaping from the Barren Plateau via Gaussian Initializations in Deep Variational Quantum Circuits

• Kaining Zhang
• Liu Liu
• Min-Hsiu Hsieh
• Dacheng Tao

Variational quantum circuits have been widely employed in quantum simulation and quantum machine learning in recent years. However, quantum circuits with random structures have poor trainability due to the exponentially vanishing gradient with respect to the circuit depth and the qubit number. This result leads to a general standpoint that deep quantum circuits would not be feasible for practical tasks. In this work, we propose an initialization strategy with theoretical guarantees for the vanishing gradient problem in general deep quantum circuits. Specifically, we prove that under proper Gaussian initialized parameters, the norm of the gradient decays at most polynomially when the qubit number and the circuit depth increase. Our theoretical results hold for both the local and the global observable cases, where the latter was believed to have vanishing gradients even for very shallow circuits. Experimental results verify our theoretical findings in quantum simulation and quantum chemistry.

ICLR Conference 2022 Conference Paper

FP-DETR: Detection Transformer Advanced by Fully Pre-training

• Wen Wang 0009
• Yang Cao 0010
• Jing Zhang 0037
• Dacheng Tao

Large-scale pre-training has proven to be effective for visual representation learning on downstream tasks, especially for improving robustness and generalization. However, the recently developed detection transformers only employ pre-training on its backbone while leaving the key component, i.e., a 12-layer transformer, being trained from scratch, which prevents the model from above benefits. This separated training paradigm is mainly caused by the discrepancy between the upstream and downstream tasks. To mitigate the issue, we propose FP-DETR, a new method that Fully Pre-Trains an encoder-only transformer and smoothly fine-tunes it for object detection via a task adapter. Inspired by the success of textual prompts in NLP, we treat query positional embeddings as visual prompts to help the model attend to the target area (prompting) and recognize the object. To this end, we propose the task adapter which leverages self-attention to model the contextual relation between object query embedding. Experiments on the challenging COCO dataset demonstrate that our FP-DETR achieves competitive performance. Moreover, it enjoys better robustness to common corruptions and generalization to small-size datasets than state-of-the-art detection transformers. Code will be made publicly available at $\url{https://github.com/encounter1997/FP-DETR}$.

TMLR Journal 2022 Journal Article

GFNet: Geometric Flow Network for 3D Point Cloud Semantic Segmentation

• Haibo Qiu
• Baosheng Yu
• Dacheng Tao

Point cloud semantic segmentation from projected views, such as range-view (RV) and bird's-eye-view (BEV), has been intensively investigated. Different views capture different information of point clouds and thus are complementary to each other. However, recent projection-based methods for point cloud semantic segmentation usually utilize a vanilla late fusion strategy for the predictions of different views, failing to explore the complementary information from a geometric perspective during the representation learning. In this paper, we introduce a geometric flow network (GFNet) to explore the geometric correspondence between different views in an align-before-fuse manner. Specifically, we devise a novel geometric flow module (GFM) to bidirectionally align and propagate the complementary information across different views according to geometric relationships under the end-to-end learning scheme. We perform extensive experiments on two widely used benchmark datasets, SemanticKITTI and nuScenes, to demonstrate the effectiveness of our GFNet for project-based point cloud semantic segmentation. Concretely, GFNet not only significantly boosts the performance of each individual view but also achieves state-of-the-art results over all existing projection-based models. Code is available at \url{https://github.com/haibo-qiu/GFNet}.

ICLR Conference 2022 Conference Paper

Huber Additive Models for Non-stationary Time Series Analysis

• Yingjie Wang 0007
• Xianrui Zhong
• Fengxiang He
• Hong Chen 0004
• Dacheng Tao

Sparse additive models have shown promising flexibility and interpretability in processing time series data. However, existing methods usually assume the time series data to be stationary and the innovation is sampled from a Gaussian distribution. Both assumptions are too stringent for heavy-tailed and non-stationary time series data that frequently arise in practice, such as finance and medical fields. To address these problems, we propose an adaptive sparse Huber additive model for robust forecasting in both non-Gaussian data and (non)stationary data. In theory, the generalization bounds of our estimator are established for both stationary and nonstationary time series data, which are independent of the widely used mixing conditions in learning theory of dependent observations. Moreover, the error bound for non-stationary time series contains a discrepancy measure for the shifts of the data distributions over time. Such a discrepancy measure can be estimated empirically and used as a penalty in our method. Experimental results on both synthetic and real-world benchmark datasets validate the effectiveness of the proposed method. The code is available at https://github.com/xianruizhong/SpHAM.

ICML Conference 2022 Conference Paper

Identity-Disentangled Adversarial Augmentation for Self-supervised Learning

• Kaiwen Yang
• Tianyi Zhou 0001
• Xinmei Tian 0001
• Dacheng Tao

Data augmentation is critical to contrastive self-supervised learning, whose goal is to distinguish a sample’s augmentations (positives) from other samples (negatives). However, strong augmentations may change the sample-identity of the positives, while weak augmentation produces easy positives/negatives leading to nearly-zero loss and ineffective learning. In this paper, we study a simple adversarial augmentation method that can modify training data to be hard positives/negatives without distorting the key information about their original identities. In particular, we decompose a sample $x$ to be its variational auto-encoder (VAE) reconstruction $G(x)$ plus the residual $R(x)=x-G(x)$, where $R(x)$ retains most identity-distinctive information due to an information-theoretic interpretation of the VAE objective. We then adversarially perturb $G(x)$ in the VAE’s bottleneck space and adds it back to the original $R(x)$ as an augmentation, which is therefore sufficiently challenging for contrastive learning and meanwhile preserves the sample identity intact. We apply this “identity-disentangled adversarial augmentation (IDAA)” to different self-supervised learning methods. On multiple benchmark datasets, IDAA consistently improves both their efficiency and generalization performance. We further show that IDAA learned on a dataset can be transferred to other datasets. Code is available at \href{https: //github. com/kai-wen-yang/IDAA}{https: //github. com/kai-wen-yang/IDAA}.

NeurIPS Conference 2022 Conference Paper

Inducing Neural Collapse in Imbalanced Learning: Do We Really Need a Learnable Classifier at the End of Deep Neural Network?

• Yibo Yang
• Shixiang Chen
• Xiangtai Li
• Liang Xie
• Zhouchen Lin
• Dacheng Tao

Modern deep neural networks for classification usually jointly learn a backbone for representation and a linear classifier to output the logit of each class. A recent study has shown a phenomenon called neural collapse that the within-class means of features and the classifier vectors converge to the vertices of a simplex equiangular tight frame (ETF) at the terminal phase of training on a balanced dataset. Since the ETF geometric structure maximally separates the pair-wise angles of all classes in the classifier, it is natural to raise the question, why do we spend an effort to learn a classifier when we know its optimal geometric structure? In this paper, we study the potential of learning a neural network for classification with the classifier randomly initialized as an ETF and fixed during training. Our analytical work based on the layer-peeled model indicates that the feature learning with a fixed ETF classifier naturally leads to the neural collapse state even when the dataset is imbalanced among classes. We further show that in this case the cross entropy (CE) loss is not necessary and can be replaced by a simple squared loss that shares the same global optimality but enjoys a better convergence property. Our experimental results show that our method is able to bring significant improvements with faster convergence on multiple imbalanced datasets.

ICLR Conference 2022 Conference Paper

Knowledge Removal in Sampling-based Bayesian Inference

• Shaopeng Fu
• Fengxiang He
• Dacheng Tao

The right to be forgotten has been legislated in many countries, but its enforcement in the AI industry would cause unbearable costs. When single data deletion requests come, companies may need to delete the whole models learned with massive resources. Existing works propose methods to remove knowledge learned from data for explicitly parameterized models, which however are not appliable to the sampling-based Bayesian inference, {\it i.e.}, Markov chain Monte Carlo (MCMC), as MCMC can only infer implicit distributions. In this paper, we propose the first machine unlearning algorithm for MCMC. We first convert the MCMC unlearning problem into an explicit optimization problem. Based on this problem conversion, an {\it MCMC influence function} is designed to provably characterize the learned knowledge from data, which then delivers the MCMC unlearning algorithm. Theoretical analysis shows that MCMC unlearning would not compromise the generalizability of the MCMC models. Experiments on Gaussian mixture models and Bayesian neural networks confirm the effectiveness of the proposed algorithm. The code is available at \url{https://github.com/fshp971/mcmc-unlearning}.

NeurIPS Conference 2022 Conference Paper

Make Sharpness-Aware Minimization Stronger: A Sparsified Perturbation Approach

• Peng Mi
• Li Shen
• Tianhe Ren
• Yiyi Zhou
• Xiaoshuai Sun
• Rongrong Ji
• Dacheng Tao

Deep neural networks often suffer from poor generalization caused by complex and non-convex loss landscapes. One of the popular solutions is Sharpness-Aware Minimization (SAM), which smooths the loss landscape via minimizing the maximized change of training loss when adding a perturbation to the weight. However, we find the indiscriminate perturbation of SAM on all parameters is suboptimal, which also results in excessive computation, ~\emph{i. e. }, double the overhead of common optimizers like Stochastic Gradient Descent~(SGD). In this paper, we propose an efficient and effective training scheme coined as Sparse SAM (SSAM), which achieves sparse perturbation by a binary mask. To obtain the sparse mask, we provide two solutions which are based on Fisher information and dynamic sparse training, respectively. In addition, we theoretically prove that SSAM can converge at the same rate as SAM, ~\emph{i. e. }, $O(\log T/\sqrt{T})$. Sparse SAM not only has the potential for training acceleration but also smooths the loss landscape effectively. Extensive experimental results on CIFAR10, CIFAR100, and ImageNet-1K confirm the superior efficiency of our method to SAM, and the performance is preserved or even better with a perturbation of merely 50\% sparsity. Code is available at \url{https: //github. com/Mi-Peng/Sparse-Sharpness-Aware-Minimization}.

IJCAI Conference 2022 Conference Paper

Penalized Proximal Policy Optimization for Safe Reinforcement Learning

• Linrui Zhang
• Li Shen
• Long Yang
• Shixiang Chen
• Xueqian Wang
• Bo Yuan
• Dacheng Tao

Safe reinforcement learning aims to learn the optimal policy while satisfying safety constraints, which is essential in real-world applications. However, current algorithms still struggle for efficient policy updates with hard constraint satisfaction. In this paper, we propose Penalized Proximal Policy Optimization (P3O), which solves the cumbersome constrained policy iteration via a single minimization of an equivalent unconstrained problem. Specifically, P3O utilizes a simple yet effective penalty approach to eliminate cost constraints and removes the trust-region constraint by the clipped surrogate objective. We theoretically prove the exactness of the penalized method with a finite penalty factor and provide a worst-case analysis for approximate error when evaluated on sample trajectories. Moreover, we extend P3O to more challenging multi-constraint and multi-agent scenarios which are less studied in previous work. Extensive experiments show that P3O outperforms state-of-the-art algorithms with respect to both reward improvement and constraint satisfaction on a set of constrained locomotive tasks.

ICLR Conference 2022 Conference Paper

Rethinking Class-Prior Estimation for Positive-Unlabeled Learning

• Yu Yao 0005
• Tongliang Liu
• Bo Han 0003
• Mingming Gong
• Gang Niu 0001
• Masashi Sugiyama
• Dacheng Tao

Given only positive (P) and unlabeled (U) data, PU learning can train a binary classifier without any negative data. It has two building blocks: PU class-prior estimation (CPE) and PU classification; the latter has been well studied while the former has received less attention. Hitherto, the distributional-assumption-free CPE methods rely on a critical assumption that the support of the positive data distribution cannot be contained in the support of the negative data distribution. If this is violated, those CPE methods will systematically overestimate the class prior; it is even worse that we cannot verify the assumption based on the data. In this paper, we rethink CPE for PU learning—can we remove the assumption to make CPE always valid? We show an affirmative answer by proposing Regrouping CPE (ReCPE) that builds an auxiliary probability distribution such that the support of the positive data distribution is never contained in the support of the negative data distribution. ReCPE can work with any CPE method by treating it as the base method. Theoretically, ReCPE does not affect its base if the assumption already holds for the original probability distribution; otherwise, it reduces the positive bias of its base. Empirically, ReCPE improves all state-of-the-art CPE methods on various datasets, implying that the assumption has indeed been violated here.

ICLR Conference 2022 Conference Paper

Robust Unlearnable Examples: Protecting Data Privacy Against Adversarial Learning

• Shaopeng Fu
• Fengxiang He
• Yang Liu 0039
• Li Shen 0008
• Dacheng Tao

The tremendous amount of accessible data in cyberspace face the risk of being unauthorized used for training deep learning models. To address this concern, methods are proposed to make data unlearnable for deep learning models by adding a type of error-minimizing noise. However, such conferred unlearnability is found fragile to adversarial training. In this paper, we design new methods to generate robust unlearnable examples that are protected from adversarial training. We first find that the vanilla error-minimizing noise, which suppresses the informative knowledge of data via minimizing the corresponding training loss, could not effectively minimize the adversarial training loss. This explains the vulnerability of error-minimizing noise in adversarial training. Based on the observation, robust error-minimizing noise is then introduced to reduce the adversarial training loss. Experiments show that the unlearnability brought by robust error-minimizing noise can effectively protect data from adversarial training in various scenarios. The code is available at \url{https://github.com/fshp971/robust-unlearnable-examples}.

IROS Conference 2022 Conference Paper

Safety Correction from Baseline: Towards the Risk-aware Policy in Robotics via Dual-agent Reinforcement Learning

• Linrui Zhang
• Zichen Yan
• Li Shen 0008
• Shoujie Li
• Xueqian Wang 0001
• Dacheng Tao

Learning a risk-aware policy is essential but rather challenging in unstructured robotic tasks. Safe reinforcement learning methods open up new possibilities to tackle this problem. However, the conservative policy updates make it intractable to achieve sufficient exploration and desirable performance in complex, sample-expensive environments. In this paper, we propose a dual-agent safe reinforcement learning strategy consisting of a baseline and a safe agent. Such a decoupled framework enables high flexibility, data efficiency and risk-awareness for RL-based control. Concretely, the baseline agent is responsible for maximizing rewards under standard RL settings. Thus, it is compatible with off-the-shelf training techniques of unconstrained optimization, exploration and exploitation. On the other hand, the safe agent mimics the baseline agent for policy improvement and learns to fulfill safety constraints via off-policy RL tuning. In contrast to training from scratch, safe policy correction requires significantly fewer interactions to obtain a near-optimal policy. The dual policies can be optimized synchronously via a shared replay buffer, or leveraging the pre-trained model or the non-learning-based controller as a fixed baseline agent. Experimental results show that our approach can learn feasible skills without prior knowledge as well as deriving risk-averse counterparts from pre-trained unsafe policies. The proposed method outperforms the state-of-the-art safe RL algorithms on difficult robot locomotion and manipulation tasks with respect to both safety constraint satisfaction and sample efficiency.

AAAI Conference 2022 Conference Paper

SASA: Semantics-Augmented Set Abstraction for Point-Based 3D Object Detection

• Chen Chen
• Zhe Chen
• Jing Zhang
• Dacheng Tao

Although point-based networks are demonstrated to be accurate for 3D point cloud modeling, they are still falling behind their voxel-based competitors in 3D detection. We observe that the prevailing set abstraction design for downsampling points may maintain too much unimportant background information that can affect feature learning for detecting objects. To tackle this issue, we propose a novel set abstraction method named Semantics-Augmented Set Abstraction (SASA). Technically, we first add a binary segmentation module as the side output to help identify foreground points. Based on the estimated point-wise foreground scores, we then propose a semantics-guided point sampling algorithm to help retain more important foreground points during downsampling. In practice, SASA shows to be effective in identifying valuable points related to foreground objects and improving feature learning for point-based 3D detection. Additionally, it is an easy-to-plug-in module and able to boost various point-based detectors, including single-stage and twostage ones. Extensive experiments on the popular KITTI and nuScenes datasets validate the superiority of SASA, lifting point-based detection models to reach comparable performance to state-of-the-art voxel-based methods. Code is available at https: //github. com/blakechen97/SASA.

ICML Conference 2022 Conference Paper

Topology-aware Generalization of Decentralized SGD

• Tongtian Zhu
• Fengxiang He
• Lan Zhang 0002
• Zhengyang Niu
• Mingli Song
• Dacheng Tao

This paper studies the algorithmic stability and generalizability of decentralized stochastic gradient descent (D-SGD). We prove that the consensus model learned by D-SGD is $\mathcal{O}{(m/N\unaryplus1/m\unaryplus\lambda^2)}$-stable in expectation in the non-convex non-smooth setting, where $N$ is the total sample size of the whole system, $m$ is the worker number, and $1\unaryminus\lambda$ is the spectral gap that measures the connectivity of the communication topology. These results then deliver an $\mathcal{O}{(1/N\unaryplus{({(m^{-1}\lambda^2)}^{\frac{\alpha}{2}}\unaryplus m^{\unaryminus\alpha})}/{N^{1\unaryminus\frac{\alpha}{2}}})}$ in-average generalization bound, which is non-vacuous even when $\lambda$ is closed to $1$, in contrast to vacuous as suggested by existing literature on the projected version of D-SGD. Our theory indicates that the generalizability of D-SGD has a positive correlation with the spectral gap, and can explain why consensus control in initial training phase can ensure better generalization. Experiments of VGG-11 and ResNet-18 on CIFAR-10, CIFAR-100 and Tiny-ImageNet justify our theory. To our best knowledge, this is the first work on the topology-aware generalization of vanilla D-SGD. Code is available at \url{https: //github. com/Raiden-Zhu/Generalization-of-DSGD}.

ICRA Conference 2022 Conference Paper

Towards Scale Consistent Monocular Visual Odometry by Learning from the Virtual World

• Sen Zhang 0006
• Jing Zhang 0037
• Dacheng Tao

Monocular visual odometry (VO) has attracted extensive research attention by providing real-time vehicle motion from cost-effective camera images. However, state-of-the-art optimization-based monocular VO methods suffer from the scale inconsistency problem for long-term predictions. Deep learning has recently been introduced to address this issue by leveraging stereo sequences or ground-truth motions in the training dataset. However, it comes at an additional cost for data collection, and such training data may not be available in all datasets. In this work, we propose VRVO, a novel framework for retrieving the absolute scale from virtual data that can be easily obtained from modern simulation environments, whereas in the real domain no stereo or ground-truth data are required in either the training or inference phases. Specifically, we first train a scale-aware disparity network using both monocular real images and stereo virtual data. The virtual-to-real domain gap is bridged by using an adversarial training strategy to map images from both domains into a shared feature space. The resulting scale-consistent disparities are then integrated with a direct VO system by constructing a virtual stereo objective that ensures the scale consistency over long trajectories. Additionally, to address the suboptimality issue caused by the separate optimization backend and the learning process, we further propose a mutual reinforcement pipeline that allows bidirectional information flow between learning and optimization, which boosts the robustness and accuracy of each other. We demonstrate the effectiveness of our framework on the KITTI and vKITTI2 datasets.

NeurIPS Conference 2022 Conference Paper

ViTPose: Simple Vision Transformer Baselines for Human Pose Estimation

• Yufei Xu
• Jing Zhang
• Qiming Zhang
• Dacheng Tao

Although no specific domain knowledge is considered in the design, plain vision transformers have shown excellent performance in visual recognition tasks. However, little effort has been made to reveal the potential of such simple structures for pose estimation tasks. In this paper, we show the surprisingly good capabilities of plain vision transformers for pose estimation from various aspects, namely simplicity in model structure, scalability in model size, flexibility in training paradigm, and transferability of knowledge between models, through a simple baseline model called ViTPose. Specifically, ViTPose employs plain and non-hierarchical vision transformers as backbones to extract features for a given person instance and a lightweight decoder for pose estimation. It can be scaled up from 100M to 1B parameters by taking the advantages of the scalable model capacity and high parallelism of transformers, setting a new Pareto front between throughput and performance. Besides, ViTPose is very flexible regarding the attention type, input resolution, pre-training and finetuning strategy, as well as dealing with multiple pose tasks. We also empirically demonstrate that the knowledge of large ViTPose models can be easily transferred to small ones via a simple knowledge token. Experimental results show that our basic ViTPose model outperforms representative methods on the challenging MS COCO Keypoint Detection benchmark, while the largest model sets a new state-of-the-art. The code and models are available at https: //github. com/ViTAE-Transformer/ViTPose.

IJCAI Conference 2021 Conference Paper

A Comprehensive Survey on Image Dehazing Based on Deep Learning

• Jie Gui
• Xiaofeng Cong
• Yuan Cao
• Wenqi Ren
• Jun Zhang
• Jing Zhang
• Dacheng Tao

The presence of haze significantly reduces the quality of images. Researchers have designed a variety of algorithms for image dehazing (ID) to restore the quality of hazy images. However, there are few studies that summarize the deep learning (DL) based dehazing technologies. In this paper, we conduct a comprehensive survey on the recent proposed dehazing methods. Firstly, we conclude the commonly used datasets, loss functions and evaluation metrics. Secondly, we group the existing researches of ID into two major categories: supervised ID and unsupervised ID. The core ideas of various influential dehazing models are introduced. Finally, the open issues for future research on ID are pointed out.

NeurIPS Conference 2021 Conference Paper

AP-10K: A Benchmark for Animal Pose Estimation in the Wild

• Hang Yu
• Yufei Xu
• Jing Zhang
• Wei Zhao
• Ziyu Guan
• Dacheng Tao

Accurate animal pose estimation is an essential step towards understanding animal behavior, and can potentially benefit many downstream applications, such as wildlife conservation. Previous works only focus on specific animals while ignoring the diversity of animal species, limiting the generalization ability. In this paper, we propose AP-10K, the first large-scale benchmark for general animal pose estimation, to facilitate the research in animal pose estimation. AP-10K consists of 10, 015 images collected and filtered from 23 animal families and 54 species following the taxonomic rank and high-quality keypoint annotations labeled and checked manually. Based on AP-10K, we benchmark representative pose estimation models on the following three tracks: (1) supervised learning for animal pose estimation, (2) cross-domain transfer learning from human pose estimation to animal pose estimation, and (3) intra- and inter-family domain generalization for unseen animals. The experimental results provide sound empirical evidence on the superiority of learning from diverse animals species in terms of both accuracy and generalization ability. It opens new directions for facilitating future research in animal pose estimation. AP-10k is publicly available at https: //github. com/AlexTheBad/AP10K.

IJCAI Conference 2021 Conference Paper

BESA: BERT-based Simulated Annealing for Adversarial Text Attacks

• Xinghao Yang
• Weifeng Liu
• Dacheng Tao
• Wei Liu

Modern Natural Language Processing (NLP) models are known immensely brittle towards text adversarial examples. Recent attack algorithms usually adopt word-level substitution strategies following a pre-computed word replacement mechanism. However, their resultant adversarial examples are still imperfect in achieving grammar correctness and semantic similarities, which is largely because of their unsuitable candidate word selections and static optimization methods. In this research, we propose BESA, a BERT-based Simulated Annealing algorithm, to address these two problems. Firstly, we leverage the BERT Masked Language Model (MLM) to generate contextual-aware candidate words to produce fluent adversarial text and avoid grammar errors. Secondly, we employ Simulated Annealing (SA) to adaptively determine the word substitution order. The SA provides sufficient word replacement options via internal simulations, with an objective to obtain both a high attack success rate and a low word substitution rate. Besides, our algorithm is able to jump out of local optima with a controlled probability, making it closer to achieve the best possible attack (i. e. , the global optima). Experiments on five popular datasets manifest the superiority of BESA compared with existing methods, including TextFooler, BAE, BERT-Attack, PWWS, and PSO.

AAAI Conference 2021 Conference Paper

Bigram and Unigram Based Text Attack via Adaptive Monotonic Heuristic Search

• Xinghao Yang
• Weifeng Liu
• James Bailey
• Dacheng Tao
• Wei Liu

Deep neural networks (DNNs) are known to be vulnerable to adversarial images, while their robustness in text classification are rarely studied. Several lines of text attack methods have been proposed in the literature, such as character-level, word-level, and sentence-level attacks. However, it is still a challenge to minimize the number of word distortions necessary to induce misclassification, while simultaneously ensuring the lexical correctness, syntactic correctness, and semantic similarity. In this paper, we propose the Bigram and Unigram based Monotonic Heuristic Search (BU-MHS) method to examine the vulnerability of deep models. Our method has three major merits. Firstly, we propose to attack text documents not only at the unigram word level but also at the bigram level to avoid producing meaningless outputs. Secondly, we propose a hybrid method to replace the input words with both their synonyms and sememe candidates, which greatly enriches potential substitutions compared to only using synonyms. Lastly, we design a search algorithm, i. e. , Monotonic Heuristic Search (MHS), to determine the priority of word replacements, aiming to reduce the modification cost in an adversarial attack. We evaluate the effectiveness of BU-MHS on IMDB, AG’s News, and Yahoo! Answers text datasets by attacking four popular DNNs models. Results show that our BU-MHS achieves the highest attack success rate by changing the smallest number of words compared with baselines.

NeurIPS Conference 2021 Conference Paper

Class-Disentanglement and Applications in Adversarial Detection and Defense

• Kaiwen Yang
• Tianyi Zhou
• Yonggang Zhang
• Xinmei Tian
• Dacheng Tao

What is the minimum necessary information required by a neural net $D(\cdot)$ from an image $x$ to accurately predict its class? Extracting such information in the input space from $x$ can allocate the areas $D(\cdot)$ mainly attending to and shed novel insights to the detection and defense of adversarial attacks. In this paper, we propose ''class-disentanglement'' that trains a variational autoencoder $G(\cdot)$ to extract this class-dependent information as $x - G(x)$ via a trade-off between reconstructing $x$ by $G(x)$ and classifying $x$ by $D(x-G(x))$, where the former competes with the latter in decomposing $x$ so the latter retains only necessary information for classification in $x-G(x)$. We apply it to both clean images and their adversarial images and discover that the perturbations generated by adversarial attacks mainly lie in the class-dependent part $x-G(x)$. The decomposition results also provide novel interpretations to classification and attack models. Inspired by these observations, we propose to conduct adversarial detection and adversarial defense respectively on $x - G(x)$ and $G(x)$, which consistently outperform the results on the original $x$. In experiments, this simple approach substantially improves the detection and defense against different types of adversarial attacks.

ICML Conference 2021 Conference Paper

Commutative Lie Group VAE for Disentanglement Learning

• Xinqi Zhu
• Chang Xu 0002
• Dacheng Tao

We view disentanglement learning as discovering an underlying structure that equivariantly reflects the factorized variations shown in data. Traditionally, such a structure is fixed to be a vector space with data variations represented by translations along individual latent dimensions. We argue this simple structure is suboptimal since it requires the model to learn to discard the properties (e. g. different scales of changes, different levels of abstractness) of data variations, which is an extra work than equivariance learning. Instead, we propose to encode the data variations with groups, a structure not only can equivariantly represent variations, but can also be adaptively optimized to preserve the properties of data variations. Considering it is hard to conduct training on group structures, we focus on Lie groups and adopt a parameterization using Lie algebra. Based on the parameterization, some disentanglement learning constraints are naturally derived. A simple model named Commutative Lie Group VAE is introduced to realize the group-based disentanglement learning. Experiments show that our model can effectively learn disentangled representations without supervision, and can achieve state-of-the-art performance without extra constraints.

IJCAI Conference 2021 Conference Paper

Deep Automatic Natural Image Matting

• Jizhizi Li
• Jing Zhang
• Dacheng Tao

Automatic image matting (AIM) refers to estimating the soft foreground from an arbitrary natural image without any auxiliary input like trimap, which is useful for image editing. Prior methods try to learn semantic features to aid the matting process while being limited to images with salient opaque foregrounds such as humans and animals. In this paper, we investigate the difficulties when extending them to natural images with salient transparent/meticulous foregrounds or non-salient foregrounds. To address the problem, a novel end-to-end matting network is proposed, which can predict a generalized trimap for any image of the above types as a unified semantic representation. Simultaneously, the learned semantic features guide the matting network to focus on the transition areas via an attention mechanism. We also construct a test set AIM-500 that contains 500 diverse natural images covering all types along with manually labeled alpha mattes, making it feasible to benchmark the generalization ability of AIM models. Results of the experiments demonstrate that our network trained on available composite matting datasets outperforms existing methods both objectively and subjectively. The source code and dataset are available at https: //github. com/JizhiziLi/AIM.

AAAI Conference 2021 Conference Paper

Deep Graph-neighbor Coherence Preserving Network for Unsupervised Cross-modal Hashing

• Jun Yu
• Hao Zhou
• Yibing Zhan
• Dacheng Tao

Unsupervised cross-modal hashing (UCMH) has become a hot topic recently. Current UCMH focuses on exploring data similarities. However, current UCMH methods calculate the similarity between two data, mainly relying on the two data’s cross-modal features. These methods suffer from inaccurate similarity problems that result in a suboptimal retrieval Hamming space, because the cross-modal features between the data are not sufficient to describe the complex data relationships, such as situations where two data have different feature representations but share the inherent concepts. In this paper, we devise a deep graph-neighbor coherence preserving network (DGCPN). Specifically, DGCPN stems from graph models and explores graph-neighbor coherence by consolidating the information between data and their neighbors. DGCPN regulates comprehensive similarity preserving losses by exploiting three types of data similarities (i. e. , the graph-neighbor coherence, the coexistent similarity, and the intra- and inter-modality consistency) and designs a half-real and half-binary optimization strategy to reduce the quantization errors during hashing. Essentially, DGCPN addresses the inaccurate similarity problem by exploring and exploiting the data’s intrinsic relationships in a graph. We conduct extensive experiments on three public UCMH datasets. The experimental results demonstrate the superiority of DGCPN, e. g. , by improving the mean average precision from 0. 722 to 0. 751 on MIRFlickr-25K using 64-bit hashing codes to retrieve texts from images. We will release the source code package and the trained model on https: //github. com/Atmegal/DGCPN.

NeurIPS Conference 2021 Conference Paper

Gauge Equivariant Transformer

• Lingshen He
• Yiming Dong
• Yisen Wang
• Dacheng Tao
• Zhouchen Lin

Attention mechanism has shown great performance and efficiency in a lot of deep learning models, in which relative position encoding plays a crucial role. However, when introducing attention to manifolds, there is no canonical local coordinate system to parameterize neighborhoods. To address this issue, we propose an equivariant transformer to make our model agnostic to the orientation of local coordinate systems (\textit{i. e. }, gauge equivariant), which employs multi-head self-attention to jointly incorporate both position-based and content-based information. To enhance expressive ability, we adopt regular field of cyclic groups as feature fields in intermediate layers, and propose a novel method to parallel transport the feature vectors in these fields. In addition, we project the position vector of each point onto its local coordinate system to disentangle the orientation of the coordinate system in ambient space (\textit{i. e. }, global coordinate system), achieving rotation invariance. To the best of our knowledge, we are the first to introduce gauge equivariance to self-attention, thus name our model Gauge Equivariant Transformer (GET), which can be efficiently implemented on triangle meshes. Extensive experiments show that GET achieves state-of-the-art performance on two common recognition tasks.

YNIMG Journal 2021 Journal Article

MRI-based Alzheimer’s disease prediction via distilling the knowledge in multi-modal data

• Hao Guan
• Chaoyue Wang
• Dacheng Tao

Mild cognitive impairment (MCI) conversion prediction, i.e., identifying MCI patients of high risks converting to Alzheimer's disease (AD), is essential for preventing or slowing the progression of AD. Although previous studies have shown that the fusion of multi-modal data can effectively improve the prediction accuracy, their applications are largely restricted by the limited availability or high cost of multi-modal data. Building an effective prediction model using only magnetic resonance imaging (MRI) remains a challenging research topic. In this work, we propose a multi-modal multi-instance distillation scheme, which aims to distill the knowledge learned from multi-modal data to an MRI-based network for MCI conversion prediction. In contrast to existing distillation algorithms, the proposed multi-instance probabilities demonstrate a superior capability of representing the complicated atrophy distributions, and can guide the MRI-based network to better explore the input MRI. To our best knowledge, this is the first study that attempts to improve an MRI-based prediction model by leveraging extra supervision distilled from multi-modal information. Experiments demonstrate the advantage of our framework, suggesting its potentials in the data-limited clinical settings.

AAAI Conference 2021 Short Paper

Multi-label Few-shot Learning with Semantic Inference (Student Abstract)

• Zhen Wang
• Yiqun Duan
• Liu Liu
• Dacheng Tao

Few-shot learning can adapt the classification model to new labels with only a few labeled examples. Previous studies mainly focused on the scenario of a single category label per example but have not effectively solved the more challenging multi-label scenario, which has exponential-sized output space and low-data. In this paper, we propose a semanticaware meta-learning model for multi-label few-shot learning. Our approach can learn and infer the semantic correlation between unseen labels and historical labels to quickly adapt multi-label tasks based on only a few examples. Specifically, features can be mapped into the semantic space via label embeddings to exploit the label correlation, thus structuring the overwhelming output space. We design a novel semantic inference mechanism for leveraging prior knowledge learned from historical labels, which will produce good generalization performance on new labels to alleviate the overfitting caused by low-data. Finally, empirical results show that the proposed method significantly outperforms the existing stateof-the-art methods on the multi-label few-shot learning tasks.

IJCAI Conference 2021 Conference Paper

One-Shot Affordance Detection

• Hongchen Luo
• Wei Zhai
• Jing Zhang
• Yang Cao
• Dacheng Tao

Affordance detection refers to identifying the potential action possibilities of objects in an image, which is an important ability for robot perception and manipulation. To empower robots with this ability in unseen scenarios, we consider the challenging one-shot affordance detection problem in this paper, i. e. , given a support image that depicts the action purpose, all objects in a scene with the common affordance should be detected. To this end, we devise a One-Shot Affordance Detection (OS-AD) network that firstly estimates the purpose and then transfers it to help detect the common affordance from all candidate images. Through collaboration learning, OS-AD can capture the common characteristics between objects having the same underlying affordance and learn a good adaptation capability for perceiving unseen affordances. Besides, we build a Purpose-driven Affordance Dataset (PAD) by collecting and labeling 4k images from 31 affordance and 72 object categories. Experimental results demonstrate the superiority of our model over previous representative ones in terms of both objective metrics and visual quality. The benchmark suite is at ProjectPage.

AAAI Conference 2021 Conference Paper

Progressive One-shot Human Parsing

• Haoyu He
• Jing Zhang
• Bhavani Thuraisingham
• Dacheng Tao

Prior human parsing models are limited to parsing humans into classes pre-defined in the training data, which is not flexible to generalize to unseen classes, e. g. , new clothing in fashion analysis. In this paper, we propose a new problem named one-shot human parsing (OSHP) that requires to parse human into an open set of reference classes defined by any single reference example. During training, only base classes defined in the training set are exposed, which can overlap with part of reference classes. In this paper, we devise a novel Progressive One-shot Parsing network (POPNet) to address two critical challenges, i. e. , testing bias and small sizes. POPNet consists of two collaborative metric learning modules named Attention Guidance Module and Nearest Centroid Module, which can learn representative prototypes for base classes and quickly transfer the ability to unseen classes during testing, thereby reducing testing bias. Moreover, POPNet adopts a progressive human parsing framework that can incorporate the learned knowledge of parent classes at the coarse granularity to help recognize the descendant classes at the fine granularity, thereby handling the small sizes issue. Experiments on the ATR-OS benchmark tailored for OSHP demonstrate POPNet outperforms other representative one-shot segmentation models by large margins and establishes a strong baseline. Source code can be found at https: //github. com/Charleshhy/One-shot-Human-Parsing.

AAAI Conference 2021 Conference Paper

SnapMix: Semantically Proportional Mixing for Augmenting Fine-grained Data

• Shaoli Huang
• Xinchao Wang
• Dacheng Tao

Data mixing augmentation has proved effective in training deep models. Recent methods mix labels mainly based on the mixture proportion of image pixels. As the main discriminative information of a fine-grained image usually resides in subtle regions, methods along this line are prone to heavy label noise in fine-grained recognition. We propose in this paper a novel scheme, termed as Semantically Proportional Mixing (SnapMix), which exploits class activation map (CAM) to lessen the label noise in augmenting fine-grained data. Snap- Mix generates the target label for a mixed image by estimating its intrinsic semantic composition, and allows for asymmetric mixing operations and ensures semantic correspondence between synthetic images and target labels. Experiments show that our method consistently outperforms existing mixed-based approaches on various datasets and under different network depths. Furthermore, by incorporating the mid-level features, the proposed SnapMix achieves top-level performance, demonstrating its potential to serve as a solid baseline for fine-grained recognition.

UAI Conference 2021 Conference Paper

Tighter Generalization Bounds for Iterative Differentially Private Learning Algorithms

• Fengxiang He
• Bohan Wang
• Dacheng Tao

This paper studies the relationship between generalization and privacy preservation of machine learning in two steps. We first establish an alignment between the two facets for any learning algorithm. We prove that $(\varepsilon, \delta)$-differential privacy implies an on-average generalization bound for a multi-sample-set learning algorithm, which further leads to a high-probability bound for any learning algorithm. We then investigate how the iterative nature shared by most learning algorithms influences privacy preservation and further generalization. Three composition theorems are proved to approximate the differential privacy of an iterative algorithm through the differential privacy of its every iteration. Integrating the above two steps, we eventually deliver generalization bounds for iterative learning algorithms. Our results are strictly tighter than the existing works. Particularly, our generalization bounds do not rely on the model size which is prohibitively large in deep learning. Experiments of MLP, VGG, and ResNet on MNIST, CIFAR-10, and CIFAR-100 are in full agreement with our theory. The theory applies to a wide spectrum of learning algorithms. In this paper, it is applied to the Gaussian mechanism as an example.

ICLR Conference 2021 Conference Paper

Understanding and Improving Lexical Choice in Non-Autoregressive Translation

• Liang Ding 0006
• Longyue Wang
• Xuebo Liu 0002
• Derek F. Wong
• Dacheng Tao
• Zhaopeng Tu

Knowledge distillation (KD) is essential for training non-autoregressive translation (NAT) models by reducing the complexity of the raw data with an autoregressive teacher model. In this study, we empirically show that as a side effect of this training, the lexical choice errors on low-frequency words are propagated to the NAT model from the teacher model. To alleviate this problem, we propose to expose the raw data to NAT models to restore the useful information of low-frequency words, which are missed in the distilled data. To this end, we introduce an extra Kullback-Leibler divergence term derived by comparing the lexical choice of NAT model and that embedded in the raw data. Experimental results across language pairs and model architectures demonstrate the effectiveness and universality of the proposed approach. Extensive analyses confirm our claim that our approach improves performance by reducing the lexical choice errors on low-frequency words. Encouragingly, our approach pushes the SOTA NAT performance on the WMT14 English-German and WMT16 Romanian-English datasets up to 27.8 and 33.8 BLEU points, respectively.

NeurIPS Conference 2021 Conference Paper

ViTAE: Vision Transformer Advanced by Exploring Intrinsic Inductive Bias

• Yufei Xu
• Qiming Zhang
• Jing Zhang
• Dacheng Tao

Transformers have shown great potential in various computer vision tasks owing to their strong capability in modeling long-range dependency using the self-attention mechanism. Nevertheless, vision transformers treat an image as 1D sequence of visual tokens, lacking an intrinsic inductive bias (IB) in modeling local visual structures and dealing with scale variance. Alternatively, they require large-scale training data and longer training schedules to learn the IB implicitly. In this paper, we propose a new Vision Transformer Advanced by Exploring intrinsic IB from convolutions, i. e. , ViTAE. Technically, ViTAE has several spatial pyramid reduction modules to downsample and embed the input image into tokens with rich multi-scale context by using multiple convolutions with different dilation rates. In this way, it acquires an intrinsic scale invariance IB and is able to learn robust feature representation for objects at various scales. Moreover, in each transformer layer, ViTAE has a convolution block in parallel to the multi-head self-attention module, whose features are fused and fed into the feed-forward network. Consequently, it has the intrinsic locality IB and is able to learn local features and global dependencies collaboratively. Experiments on ImageNet as well as downstream tasks prove the superiority of ViTAE over the baseline transformer and concurrent works. Source code and pretrained models will be available at https: //github. com/Annbless/ViTAE.

NeurIPS Conference 2020 Conference Paper

Auto Learning Attention

• Benteng Ma
• Jing Zhang
• Yong Xia
• Dacheng Tao

Attention modules have been demonstrated effective in strengthening the representation ability of a neural network via reweighting spatial or channel features or stacking both operations sequentially. However, designing the structures of different attention operations requires a bulk of computation and extensive expertise. In this paper, we devise an Auto Learning Attention (AutoLA) method, which is the first attempt on automatic attention design. Specifically, we define a novel attention module named high order group attention (HOGA) as a directed acyclic graph (DAG) where each group represents a node, and each edge represents an operation of heterogeneous attentions. A typical HOGA architecture can be searched automatically via the differential AutoLA method within 1 GPU day using the ResNet-20 backbone on CIFAR10. Further, the searched attention module can generalize to various backbones as a plug-and-play component and outperforms popular manually designed channel and spatial attentions for many vision tasks, including image classification on CIFAR100 and ImageNet, object detection and human keypoint detection on COCO dataset. The code will be released.

IJCAI Conference 2020 Conference Paper

Collaborative Learning of Depth Estimation, Visual Odometry and Camera Relocalization from Monocular Videos

• Haimei Zhao
• Wei Bian
• Bo Yuan
• Dacheng Tao

Scene perceiving and understanding tasks including depth estimation, visual odometry (VO) and camera relocalization are fundamental for applications such as autonomous driving, robots and drones. Driven by the power of deep learning, significant progress has been achieved on individual tasks but the rich correlations among the three tasks are largely neglected. In previous studies, VO is generally accurate in local scope yet suffers from drift in long distances. By contrast, camera relocalization performs well in the global sense but lacks local precision. We argue that these two tasks should be strategically combined to leverage the complementary advantages, and be further improved by exploiting the 3D geometric information from depth data, which is also beneficial for depth estimation in turn. Therefore, we present a collaborative learning framework, consisting of DepthNet, LocalPoseNet and GlobalPoseNet with a joint optimization loss to estimate depth, VO and camera localization unitedly. Moreover, the Geometric Attention Guidance Model is introduced to exploit the geometric relevance among three branches during learning. Extensive experiments demonstrate that the joint learning scheme is useful for all tasks and our method outperforms current state-of-the-art techniques in depth estimation and camera relocalization with highly competitive performance in VO.

ICML Conference 2020 Conference Paper

Deep Streaming Label Learning

• Zhen Wang 0030
• Liu Liu 0014
• Dacheng Tao

In multi-label learning, each instance can be associated with multiple and non-exclusive labels. Previous studies assume that all the labels in the learning process are fixed and static; however, they ignore the fact that the labels will emerge continuously in changing environments. In order to fill in these research gaps, we propose a novel deep neural network (DNN) based framework, Deep Streaming Label Learning (DSLL), to classify instances with newly emerged labels effectively. DSLL can explore and incorporate the knowledge from past labels and historical models to understand and develop emerging new labels. DSLL consists of three components: 1) a streaming label mapping to extract deep relationships between new labels and past labels with a novel label-correlation aware loss; 2) a streaming feature distillation propagating feature-level knowledge from the historical model to a new model; 3) a senior student network to model new labels with the help of knowledge learned from the past. Theoretically, we prove that DSLL admits tight generalization error bounds for new labels in the DNN framework. Experimentally, extensive empirical results show that the proposed method performs significantly better than the existing state-of-the-art multi-label learning methods to handle the continually emerging new labels.

NeurIPS Conference 2020 Conference Paper

Domain Generalization via Entropy Regularization

• Shanshan Zhao
• Mingming Gong
• Tongliang Liu
• Huan Fu
• Dacheng Tao

Domain generalization aims to learn from multiple source domains a predictive model that can generalize to unseen target domains. One essential problem in domain generalization is to learn discriminative domain-invariant features. To arrive at this, some methods introduce a domain discriminator through adversarial learning to match the feature distributions in multiple source domains. However, adversarial training can only guarantee that the learned features have invariant marginal distributions, while the invariance of conditional distributions is more important for prediction in new domains. To ensure the conditional invariance of learned features, we propose an entropy regularization term that measures the dependency between the learned features and the class labels. Combined with the typical task-related loss, e. g. , cross-entropy loss for classification, and adversarial loss for domain discrimination, our overall objective is guaranteed to learn conditional-invariant features across all source domains and thus can learn classifiers with better generalization capabilities. We demonstrate the effectiveness of our method through comparison with state-of-the-art methods on both simulated and real-world datasets. Code is available at: https: //github. com/sshan-zhao/DG via ER.

AAAI Conference 2020 Conference Paper

Grapy-ML: Graph Pyramid Mutual Learning for Cross-Dataset Human Parsing

• Haoyu He
• Jing Zhang
• Qiming Zhang
• Dacheng Tao

Human parsing, or human body part semantic segmentation, has been an active research topic due to its wide potential applications. In this paper, we propose a novel GRAph PYramid Mutual Learning (Grapy-ML) method to address the cross-dataset human parsing problem, where the annotations are at different granularities. Starting from the prior knowledge of the human body hierarchical structure, we devise a graph pyramid module (GPM) by stacking three levels of graph structures from coarse granularity to fine granularity subsequently. At each level, GPM utilizes the selfattention mechanism to model the correlations between context nodes. Then, it adopts a top-down mechanism to progressively refine the hierarchical features through all the levels. GPM also enables efficient mutual learning. Specifically, the network weights of the first two levels are shared to exchange the learned coarse-granularity information across different datasets. By making use of the multi-granularity labels, Grapy-ML learns a more discriminative feature representation and achieves state-of-the-art performance, which is demonstrated by extensive experiments on the three popular benchmarks, e. g. CIHP dataset. The source code is publicly available at https: //github. com/Charleshhy/Grapy-ML.

NeurIPS Conference 2020 Conference Paper

Hard Example Generation by Texture Synthesis for Cross-domain Shape Similarity Learning

• Huan Fu
• Shunming Li
• Rongfei Jia
• Mingming Gong
• Binqiang Zhao
• Dacheng Tao

Image-based 3D shape retrieval (IBSR) aims to find the corresponding 3D shape of a given 2D image from a large 3D shape database. The common routine is to map 2D images and 3D shapes into an embedding space and define (or learn) a shape similarity measure. While metric learning with some adaptation techniques seems to be a natural solution to shape similarity learning, the performance is often unsatisfactory for fine-grained shape retrieval. In the paper, we identify the source of the poor performance and propose a practical solution to this problem. We find that the shape difference between a negative pair is entangled with the texture gap, making metric learning ineffective in pushing away negative pairs. To tackle this issue, we develop a geometry-focused multi-view metric learning framework empowered by texture synthesis. The synthesis of textures for 3D shape models creates hard triplets, which suppress the adverse effects of rich texture in 2D images, thereby push the network to focus more on discovering geometric characteristics. Our approach shows state-of-the-art performance on a recently released large-scale 3D-FUTURE [1] repository, as well as three widely studied benchmarks, including Pix3D [2], Stanford Cars [3], and Comp Cars [4]. Codes will be made publicly available at: https: //github. com/3D-FRONT-FUTURE/IBSR-texture.

ICML Conference 2020 Conference Paper

Label-Noise Robust Domain Adaptation

• Xiyu Yu
• Tongliang Liu
• Mingming Gong
• Kun Zhang 0001
• Kayhan Batmanghelich
• Dacheng Tao

Domain adaptation aims to correct the classifiers when faced with distribution shift between source (training) and target (test) domains. State-of-the-art domain adaptation methods make use of deep networks to extract domain-invariant representations. However, existing methods assume that all the instances in the source domain are correctly labeled; while in reality, it is unsurprising that we may obtain a source domain with noisy labels. In this paper, we are the first to comprehensively investigate how label noise could adversely affect existing domain adaptation methods in various scenarios. Further, we theoretically prove that there exists a method that can essentially reduce the side-effect of noisy source labels in domain adaptation. Specifically, focusing on the generalized target shift scenario, where both label distribution $P_Y$ and the class-conditional distribution $P_{X|Y}$ can change, we discover that the denoising Conditional Invariant Component (DCIC) framework can provably ensures (1) extracting invariant representations given examples with noisy labels in the source domain and unlabeled examples in the target domain and (2) estimating the label distribution in the target domain with no bias. Experimental results on both synthetic and real-world data verify the effectiveness of the proposed method.

ICML Conference 2020 Conference Paper

Learning with Bounded Instance and Label-dependent Label Noise

• Jiacheng Cheng
• Tongliang Liu
• Kotagiri Ramamohanarao
• Dacheng Tao

Instance- and Label-dependent label Noise (ILN) widely exists in real-world datasets but has been rarely studied. In this paper, we focus on Bounded Instance- and Label-dependent label Noise (BILN), a particular case of ILN where the label noise rates—the probabilities that the true labels of examples flip into the corrupted ones—have upper bound less than $1$. Specifically, we introduce the concept of distilled examples, i. e. examples whose labels are identical with the labels assigned for them by the Bayes optimal classifier, and prove that under certain conditions classifiers learnt on distilled examples will converge to the Bayes optimal classifier. Inspired by the idea of learning with distilled examples, we then propose a learning algorithm with theoretical guarantees for its robustness to BILN. At last, empirical evaluations on both synthetic and real-world datasets show effectiveness of our algorithm in learning with BILN.

ICML Conference 2020 Conference Paper

LTF: A Label Transformation Framework for Correcting Label Shift

• Jiaxian Guo
• Mingming Gong
• Tongliang Liu
• Kun Zhang 0001
• Dacheng Tao

Distribution shift is a major obstacle to the deployment of current deep learning models on real-world problems. Let $Y$ be the class label and $X$ the features. We focus on one type of distribution shift, \emph{ label shift}, where the label marginal distribution $P_Y$ changes but the conditional distribution $P_{X|Y}$ does not. Most existing methods estimate the density ratio between the source- and target-domain label distributions by density matching. However, these methods are either computationally infeasible for large-scale data or restricted to shift correction for discrete labels. In this paper, we propose an end-to-end Label Transformation Framework (LTF) for correcting label shift, which implicitly models the shift of $P_Y$ and the conditional distribution $P_{X|Y}$ using neural networks. Thanks to the flexibility of deep networks, our framework can handle continuous, discrete, and even multi-dimensional labels in a unified way and is scalable to large data. Moreover, for high dimensional $X$, such as images, we find that the redundant information in $X$ severely degrades the estimation accuracy. To remedy this issue, we propose to match the distribution implied by our generative model and the target-domain distribution in a low-dimensional feature space that discards information irrelevant to $Y$. Both theoretical and empirical studies demonstrate the superiority of our method over previous approaches.

NeurIPS Conference 2020 Conference Paper

Part-dependent Label Noise: Towards Instance-dependent Label Noise

• Xiaobo Xia
• Tongliang Liu
• Bo Han
• Nannan Wang
• Mingming Gong
• Haifeng Liu
• Gang Niu
• Dacheng Tao

Learning with the \textit{instance-dependent} label noise is challenging, because it is hard to model such real-world noise. Note that there are psychological and physiological evidences showing that we humans perceive instances by decomposing them into parts. Annotators are therefore more likely to annotate instances based on the parts rather than the whole instances, where a wrong mapping from parts to classes may cause the instance-dependent label noise. Motivated by this human cognition, in this paper, we approximate the instance-dependent label noise by exploiting \textit{part-dependent} label noise. Specifically, since instances can be approximately reconstructed by a combination of parts, we approximate the instance-dependent \textit{transition matrix} for an instance by a combination of the transition matrices for the parts of the instance. The transition matrices for parts can be learned by exploiting anchor points (i. e. , data points that belong to a specific class almost surely). Empirical evaluations on synthetic and real-world datasets demonstrate our method is superior to the state-of-the-art approaches for learning from the instance-dependent label noise.

ICLR Conference 2020 Conference Paper

Piecewise linear activations substantially shape the loss surfaces of neural networks

• Fengxiang He
• Bohan Wang
• Dacheng Tao

Understanding the loss surface of a neural network is fundamentally important to the understanding of deep learning. This paper presents how piecewise linear activation functions substantially shape the loss surfaces of neural networks. We first prove that {\it the loss surfaces of many neural networks have infinite spurious local minima} which are defined as the local minima with higher empirical risks than the global minima. Our result demonstrates that the networks with piecewise linear activations possess substantial differences to the well-studied linear neural networks. This result holds for any neural network with arbitrary depth and arbitrary piecewise linear activation functions (excluding linear functions) under most loss functions in practice. Essentially, the underlying assumptions are consistent with most practical circumstances where the output layer is narrower than any hidden layer. In addition, the loss surface of a neural network with piecewise linear activations is partitioned into multiple smooth and multilinear cells by nondifferentiable boundaries. The constructed spurious local minima are concentrated in one cell as a valley: they are connected with each other by a continuous path, on which empirical risk is invariant. Further for one-hidden-layer networks, we prove that all local minima in a cell constitute an equivalence class; they are concentrated in a valley; and they are all global minima in the cell.

AIJ Journal 2020 Journal Article

Robust learning with imperfect privileged information

• Xue Li
• Bo Du
• Chang Xu
• Yipeng Zhang
• Lefei Zhang
• Dacheng Tao

In the learning using privileged information (LUPI) paradigm, example data cannot always be clean, while the gathered privileged information can be imperfect in practice. Here, imperfect privileged information can refer to auxiliary information that is not always accurate or perturbed by noise, or alternatively to incomplete privileged information, where privileged information is only available for part of the training data. Because of the lack of clear strategies for handling noise in example data and imperfect privileged information, existing learning using privileged information (LUPI) methods may encounter serious issues. Accordingly, in this paper, we propose a Robust SVM+ method to tackle imperfect data in LUPI. In order to make the SVM+ model robust to noise in example data and privileged information, Robust SVM+ maximizes the lower bound of the perturbations that may influence the judgement based on a rigorous theoretical analysis. Moreover, in order to deal with the incomplete privileged information, we use the available privileged information to help us in approximating the missing privileged information of training data. The optimization problem of the proposed method can be efficiently solved by employing a two-step alternating optimization strategy, based on iteratively deploying off-the-shelf quadratic programming solvers and the alternating direction method of multipliers (ADMM) technique. Comprehensive experiments on real-world datasets demonstrate the effectiveness of the proposed Robust SVM+ method in handling imperfect privileged information.

NeurIPS Conference 2020 Conference Paper

SCOP: Scientific Control for Reliable Neural Network Pruning

• Yehui Tang
• Yunhe Wang
• Yixing Xu
• Dacheng Tao
• Chunjing Xu
• Chao Xu
• Chang Xu

This paper proposes a reliable neural network pruning algorithm by setting up a scientific control. Existing pruning methods have developed various hypotheses to approximate the importance of filters to the network and then execute filter pruning accordingly. To increase the reliability of the results, we prefer to have a more rigorous research design by including a scientific control group as an essential part to minimize the effect of all factors except the association between the filter and expected network output. Acting as a control group, knockoff feature is generated to mimic the feature map produced by the network filter, but they are conditionally independent of the example label given the real feature map. We theoretically suggest that the knockoff condition can be approximately preserved given the information propagation of network layers. Besides the real feature map on an intermediate layer, the corresponding knockoff feature is brought in as another auxiliary input signal for the subsequent layers. Redundant filters can be discovered in the adversarial process of different features. Through experiments, we demonstrate the superiority of the proposed algorithm over state-of-the-art methods. For example, our method can reduce 57. 8% parameters and 60. 2% FLOPs of ResNet-101 with only 0. 01% top-1 accuracy loss on ImageNet.

NeurIPS Conference 2020 Conference Paper

Searching for Low-Bit Weights in Quantized Neural Networks

• Zhaohui Yang
• Yunhe Wang
• Kai Han
• Chunjing Xu
• Chao Xu
• Dacheng Tao
• Chang Xu

Quantized neural networks with low-bit weights and activations are attractive for developing AI accelerators. However, the quantization functions used in most conventional quantization methods are non-differentiable, which increases the optimization difficulty of quantized networks. Compared with full-precision parameters (\emph{i. e. }, 32-bit floating numbers), low-bit values are selected from a much smaller set. For example, there are only 16 possibilities in 4-bit space. Thus, we present to regard the discrete weights in an arbitrary quantized neural network as searchable variables, and utilize a differential method to search them accurately. In particular, each weight is represented as a probability distribution over the discrete value set. The probabilities are optimized during training and the values with the highest probability are selected to establish the desired quantized network. Experimental results on benchmarks demonstrate that the proposed method is able to produce quantized neural networks with higher performance over the state-of-the-arts on both image classification and super-resolution tasks.

ICLR Conference 2020 Conference Paper

Understanding Generalization in Recurrent Neural Networks

• Zhuozhuo Tu
• Fengxiang He
• Dacheng Tao

In this work, we develop the theory for analyzing the generalization performance of recurrent neural networks. We first present a new generalization bound for recurrent neural networks based on matrix 1-norm and Fisher-Rao norm. The definition of Fisher-Rao norm relies on a structural lemma about the gradient of RNNs. This new generalization bound assumes that the covariance matrix of the input data is positive definite, which might limit its use in practice. To address this issue, we propose to add random noise to the input data and prove a generalization bound for training with random noise, which is an extension of the former one. Compared with existing results, our generalization bounds have no explicit dependency on the size of networks. We also discover that Fisher-Rao norm for RNNs can be interpreted as a measure of gradient, and incorporating this gradient measure not only can tighten the bound, but allows us to build a relationship between generalization and trainability. Based on the bound, we theoretically analyze the effect of covariance of features on generalization of RNNs and discuss how weight decay and gradient clipping in the training can help improve generalization.

NeurIPS Conference 2020 Conference Paper

Video Frame Interpolation without Temporal Priors

• Youjian Zhang
• Chaoyue Wang
• Dacheng Tao

Video frame interpolation, which aims to synthesize non-exist intermediate frames in a video sequence, is an important research topic in computer vision. Existing video frame interpolation methods have achieved remarkable results under specific assumptions, such as instant or known exposure time. However, in complicated real-world situations, the temporal priors of videos, i. e. frames per second (FPS) and frame exposure time, may vary from different camera sensors. When test videos are taken under different exposure settings from training ones, the interpolated frames will suffer significant misalignment problems. In this work, we solve the video frame interpolation problem in a general situation, where input frames can be acquired under uncertain exposure (and interval) time. Unlike previous methods that can only be applied to a specific temporal prior, we derive a general curvilinear motion trajectory formula from four consecutive sharp frames or two consecutive blurry frames without temporal priors. Moreover, utilizing constraints within adjacent motion trajectories, we devise a novel optical flow refinement strategy for better interpolation results. Finally, experiments demonstrate that one well-trained model is enough for synthesizing high-quality slow-motion videos under complicated real-world situations. Codes are available on https: //github. com/yjzhang96/UTI-VFI.

AAAI Conference 2019 Conference Paper

ActivityNet-QA: A Dataset for Understanding Complex Web Videos via Question Answering

• Zhou Yu
• Dejing Xu
• Jun Yu
• Ting Yu
• Zhou Zhao
• Yueting Zhuang
• Dacheng Tao

Recent developments in modeling language and vision have been successfully applied to image question answering. It is both crucial and natural to extend this research direction to the video domain for video question answering (VideoQA). Compared to the image domain where large scale and fully annotated benchmark datasets exists, VideoQA datasets are limited to small scale and are automatically generated, etc. These limitations restrict their applicability in practice. Here we introduce ActivityNet-QA, a fully annotated and large scale VideoQA dataset. The dataset consists of 58, 000 QA pairs on 5, 800 complex web videos derived from the popular ActivityNet dataset. We present a statistical analysis of our ActivityNet-QA dataset and conduct extensive experiments on it by comparing existing VideoQA baselines. Moreover, we explore various video representation strategies to improve VideoQA performance, especially for long videos.

NeurIPS Conference 2019 Conference Paper

Category Anchor-Guided Unsupervised Domain Adaptation for Semantic Segmentation

• Qiming Zhang
• Jing Zhang
• Wei Liu
• Dacheng Tao

Unsupervised domain adaptation (UDA) aims to enhance the generalization capability of a certain model from a source domain to a target domain. UDA is of particular significance since no extra effort is devoted to annotating target domain samples. However, the different data distributions in the two domains, or \emph{domain shift/discrepancy}, inevitably compromise the UDA performance. Although there has been a progress in matching the marginal distributions between two domains, the classifier favors the source domain features and makes incorrect predictions on the target domain due to category-agnostic feature alignment. In this paper, we propose a novel category anchor-guided (CAG) UDA model for semantic segmentation, which explicitly enforces category-aware feature alignment to learn shared discriminative features and classifiers simultaneously. First, the category-wise centroids of the source domain features are used as guided anchors to identify the active features in the target domain and also assign them pseudo-labels. Then, we leverage an anchor-based pixel-level distance loss and a discriminative loss to drive the intra-category features closer and the inter-category features further apart, respectively. Finally, we devise a stagewise training mechanism to reduce the error accumulation and adapt the proposed model progressively. Experiments on both the GTA5$\rightarrow $Cityscapes and SYNTHIA$\rightarrow $Cityscapes scenarios demonstrate the superiority of our CAG-UDA model over the state-of-the-art methods. The code is available at \url{https: //github. com/RogerZhangzz/CAG\_UDA}.

NeurIPS Conference 2019 Conference Paper

Control Batch Size and Learning Rate to Generalize Well: Theoretical and Empirical Evidence

• Fengxiang He
• Tongliang Liu
• Dacheng Tao

Deep neural networks have received dramatic success based on the optimization method of stochastic gradient descent (SGD). However, it is still not clear how to tune hyper-parameters, especially batch size and learning rate, to ensure good generalization. This paper reports both theoretical and empirical evidence of a training strategy that we should control the ratio of batch size to learning rate not too large to achieve a good generalization ability. Specifically, we prove a PAC-Bayes generalization bound for neural networks trained by SGD, which has a positive correlation with the ratio of batch size to learning rate. This correlation builds the theoretical foundation of the training strategy. Furthermore, we conduct a large-scale experiment to verify the correlation and training strategy. We trained 1, 600 models based on architectures ResNet-110, and VGG-19 with datasets CIFAR-10 and CIFAR-100 while strictly control unrelated variables. Accuracies on the test sets are collected for the evaluation. Spearman's rank-order correlation coefficients and the corresponding $p$ values on 164 groups of the collected data demonstrate that the correlation is statistically significant, which fully supports the training strategy.

TIST Journal 2019 Journal Article

Deep Multi-scale Discriminative Networks for Double JPEG Compression Forensics

• Cheng Deng
• Zhao Li
• Xinbo Gao
• Dacheng Tao

As JPEG is the most widely used image format, the importance of tampering detection for JPEG images in blind forensics is self-evident. In this area, extracting effective statistical characteristics from a JPEG image for classification remains a challenge. Effective features are designed manually in traditional methods, suggesting that extensive labor-consuming research and derivation is required. In this article, we propose a novel image tampering detection method based on deep multi-scale discriminative networks (MSD-Nets). The multi-scale module is designed to automatically extract multiple features from the discrete cosine transform (DCT) coefficient histograms of the JPEG image. This module can capture the characteristic information in different scale spaces. In addition, a discriminative module is also utilized to improve the detection effect of the networks in those difficult situations when the first compression quality ( QF 1) is higher than the second one ( QF 2). A special network in this module is designed to distinguish the small statistical difference between authentic and tampered regions in these cases. Finally, a probability map can be obtained and the specific tampering area is located using the last classification results. Extensive experiments demonstrate the superiority of our proposed method in both quantitative and qualitative metrics when compared with state-of-the-art approaches.

NeurIPS Conference 2019 Conference Paper

Learn, Imagine and Create: Text-to-Image Generation from Prior Knowledge

• Tingting Qiao
• Jing Zhang
• Duanqing Xu
• Dacheng Tao

Text-to-image generation, i. e. generating an image given a text description, is a very challenging task due to the significant semantic gap between the two domains. Humans, however, tackle this problem intelligently. We learn from diverse objects to form a solid prior about semantics, textures, colors, shapes, and layouts. Given a text description, we immediately imagine an overall visual impression using this prior and, based on this, we draw a picture by progressively adding more and more details. In this paper, and inspired by this process, we propose a novel text-to-image method called LeicaGAN to combine the above three phases in a unified framework. First, we formulate the multiple priors learning phase as a textual-visual co-embedding (TVE) comprising a text-image encoder for learning semantic, texture, and color priors and a text-mask encoder for learning shape and layout priors. Then, we formulate the imagination phase as multiple priors aggregation (MPA) by combining these complementary priors and adding noise for diversity. Lastly, we formulate the creation phase by using a cascaded attentive generator (CAG) to progressively draw a picture from coarse to fine. We leverage adversarial learning for LeicaGAN to enforce semantic consistency and visual realism. Thorough experiments on two public benchmark datasets demonstrate LeicaGAN's superiority over the baseline method. Code has been made available at https: //github. com/qiaott/LeicaGAN.

NeurIPS Conference 2019 Conference Paper

Learning from Bad Data via Generation

• Tianyu Guo
• Chang Xu
• Boxin Shi
• Chao Xu
• Dacheng Tao

Bad training data would challenge the learning model from understanding the underlying data-generating scheme, which then increases the difficulty in achieving satisfactory performance on unseen test data. We suppose the real data distribution lies in a distribution set supported by the empirical distribution of bad data. A worst-case formulation can be developed over this distribution set, and then be interpreted as a generation task in an adversarial manner. The connections and differences between GANs and our framework have been thoroughly discussed. We further theoretically show the influence of this generation task on learning from bad data and reveal its connection with a data-dependent regularization. Given different distance measures (\eg, Wasserstein distance or JS divergence) of distributions, we can derive different objective functions for the problem. Experimental results on different kinds of bad training data demonstrate the necessity and effectiveness of the proposed method.

NeurIPS Conference 2019 Conference Paper

LIIR: Learning Individual Intrinsic Reward in Multi-Agent Reinforcement Learning

• Yali Du
• Lei Han
• Meng Fang
• Ji Liu
• Tianhong Dai
• Dacheng Tao

A great challenge in cooperative decentralized multi-agent reinforcement learning (MARL) is generating diversified behaviors for each individual agent when receiving only a team reward. Prior studies have paid much effort on reward shaping or designing a centralized critic that can discriminatively credit the agents. In this paper, we propose to merge the two directions and learn each agent an intrinsic reward function which diversely stimulates the agents at each time step. Specifically, the intrinsic reward for a specific agent will be involved in computing a distinct proxy critic for the agent to direct the updating of its individual policy. Meanwhile, the parameterized intrinsic reward function will be updated towards maximizing the expected accumulated team reward from the environment so that the objective is consistent with the original MARL problem. The proposed method is referred to as learning individual intrinsic reward (LIIR) in MARL. We compare LIIR with a number of state-of-the-art MARL methods on battle games in StarCraft II. The results demonstrate the effectiveness of LIIR, and we show LIIR can assign each individual agent an insightful intrinsic reward per time step.

NeurIPS Conference 2019 Conference Paper

Likelihood-Free Overcomplete ICA and Applications In Causal Discovery

• Chenwei DING
• Mingming Gong
• Kun Zhang
• Dacheng Tao

Causal discovery witnessed significant progress over the past decades. In particular, many recent causal discovery methods make use of independent, non-Gaussian noise to achieve identifiability of the causal models. Existence of hidden direct common causes, or confounders, generally makes causal discovery more difficult; whenever they are present, the corresponding causal discovery algorithms can be seen as extensions of overcomplete independent component analysis (OICA). However, existing OICA algorithms usually make strong parametric assumptions on the distribution of independent components, which may be violated on real data, leading to sub-optimal or even wrong solutions. In addition, existing OICA algorithms rely on the Expectation Maximization (EM) procedure that requires computationally expensive inference of the posterior distribution of independent components. To tackle these problems, we present a Likelihood-Free Overcomplete ICA algorithm (LFOICA) that estimates the mixing matrix directly by back-propagation without any explicit assumptions on the density function of independent components. Thanks to its computational efficiency, the proposed method makes a number of causal discovery procedures much more practically feasible. For illustrative purposes, we demonstrate the computational efficiency and efficacy of our method in two causal discovery tasks on both synthetic and real data.

TIST Journal 2019 Journal Article

Multi-Modal Curriculum Learning over Graphs

• Chen Gong
• Jian Yang
• Dacheng Tao

Curriculum Learning (CL) is a recently proposed learning paradigm that aims to achieve satisfactory performance by properly organizing the learning sequence from simple curriculum examples to more difficult ones. Up to now, few works have been done to explore CL for the data with graph structure. Therefore, this article proposes a novel CL algorithm that can be utilized to guide the Label Propagation (LP) over graphs, of which the target is to “learn” the labels of unlabeled examples on the graphs. Specifically, we assume that different unlabeled examples have different levels of difficulty for propagation, and their label learning should follow a simple-to-difficult sequence with the updated curricula. Furthermore, considering that the practical data are often characterized by multiple modalities, every modality in our method is associated with a “teacher” that not only evaluates the difficulties of examples from its own viewpoint, but also cooperates with other teachers to generate the overall simplest curriculum examples for propagation. By taking the curriculums suggested by the teachers as a whole, the common preference (i.e., commonality) of teachers on selecting the simplest examples can be discovered by a row-sparse matrix, and their distinct opinions (i.e., individuality) are captured by a sparse noise matrix. As a result, an accurate curriculum sequence can be established and the propagation quality can thus be improved. Theoretically, we prove that the propagation risk bound is closely related to the examples’ difficulty information, and empirically, we show that our method can generate higher accuracy than the state-of-the-art CL approach and LP algorithms on various multi-modal tasks.

AAAI Conference 2019 Conference Paper

Perceptual-Sensitive GAN for Generating Adversarial Patches

• Aishan Liu
• Xianglong Liu
• Jiaxin Fan
• Yuqing Ma
• Anlan Zhang
• Huiyuan Xie
• Dacheng Tao

Deep neural networks (DNNs) are vulnerable to adversarial examples where inputs with imperceptible perturbations mislead DNNs to incorrect results. Recently, adversarial patch, with noise confined to a small and localized patch, emerged for its easy accessibility in real-world. However, existing attack strategies are still far from generating visually natural patches with strong attacking ability, since they often ignore the perceptual sensitivity of the attacked network to the adversarial patch, including both the correlations with the image context and the visual attention. To address this problem, this paper proposes a perceptual-sensitive generative adversarial network (PS-GAN) that can simultaneously enhance the visual fidelity and the attacking ability for the adversarial patch. To improve the visual fidelity, we treat the patch generation as a patch-to-patch translation via an adversarial process, feeding any types of seed patch and outputting the similar adversarial patch with high perceptual correlation with the attacked image. To further enhance the attacking ability, an attention mechanism coupled with adversarial generation is introduced to predict the critical attacking areas for placing the patches, which can help producing more realistic and aggressive patches. Extensive experiments under semi-whitebox and black-box settings on two large-scale datasets GTSRB and ImageNet demonstrate that the proposed PS-GAN outperforms state-of-the-art adversarial patch attack methods.

NeurIPS Conference 2019 Conference Paper

Positive-Unlabeled Compression on the Cloud

• Yixing Xu
• Yunhe Wang
• Hanting Chen
• Kai Han
• Chunjing Xu
• Dacheng Tao
• Chang Xu

Many attempts have been done to extend the great success of convolutional neural networks (CNNs) achieved on high-end GPU servers to portable devices such as smart phones. Providing compression and acceleration service of deep learning models on the cloud is therefore of significance and is attractive for end users. However, existing network compression and acceleration approaches usually fine-tuning the svelte model by requesting the entire original training data (e. g. ImageNet), which could be more cumbersome than the network itself and cannot be easily uploaded to the cloud. In this paper, we present a novel positive-unlabeled (PU) setting for addressing this problem. In practice, only a small portion of the original training set is required as positive examples and more useful training examples can be obtained from the massive unlabeled data on the cloud through a PU classifier with an attention based multi-scale feature extractor. We further introduce a robust knowledge distillation (RKD) scheme to deal with the class imbalance problem of these newly augmented training examples. The superiority of the proposed method is verified through experiments conducted on the benchmark models and datasets. We can use only 8% of uniformly selected data from the ImageNet to obtain an efficient model with comparable performance to the baseline ResNet-34.

AAAI Conference 2019 Conference Paper

Smooth Deep Image Generator from Noises

• Tianyu Guo
• Chang Xu
• Boxin Shi
• Chao Xu
• Dacheng Tao

Generative Adversarial Networks (GANs) have demonstrated a strong ability to fit complex distributions since they were presented, especially in the field of generating natural images. Linear interpolation in the noise space produces a continuously changing in the image space, which is an impressive property of GANs. However, there is no special consideration on this property in the objective function of GANs or its derived models. This paper analyzes the perturbation on the input of the generator and its influence on the generated images. A smooth generator is then developed by investigating the tolerable input perturbation. We further integrate this smooth generator with a gradient penalized discriminator, and design smooth GAN that generates stable and high-quality images. Experiments on real-world image datasets demonstrate the necessity of studying smooth generator and the effectiveness of the proposed algorithm.

IJCAI Conference 2019 Conference Paper

SPAGAN: Shortest Path Graph Attention Network

• Yiding Yang
• Xinchao Wang
• Mingli Song
• Junsong Yuan
• Dacheng Tao

Graph convolutional networks (GCN) have recently demonstrated their potential in analyzing non-grid structure data that can be represented as graphs. The core idea is to encode the local topology of a graph, via convolutions, into the feature of a center node. In this paper, we propose a novel GCN model, which we term as Shortest Path Graph Attention Network (SPAGAN). Unlike conventional GCN models that carry out node-based attentions, on either first-order neighbors or random higher-order ones, the proposed SPAGAN conducts path-based attention that explicitly accounts for the influence of a sequence of nodes yielding the minimum cost, or shortest path, between the center node and its higher-order neighbors. SPAGAN therefore allows for a more informative and intact exploration of the graph structure and further the more effective aggregation of information from distant neighbors, as compared to node-based GCN methods. We test SPAGAN for the downstream classification task on several standard datasets, and achieve performances superior to the state of the art.

NeurIPS Conference 2019 Conference Paper

Theoretical Analysis of Adversarial Learning: A Minimax Approach

• Zhuozhuo Tu
• Jingwei Zhang
• Dacheng Tao

In this paper, we propose a general theoretical method for analyzing the risk bound in the presence of adversaries. Specifically, we try to fit the adversarial learning problem into the minimax framework. We first show that the original adversarial learning problem can be transformed into a minimax statistical learning problem by introducing a transport map between distributions. Then, we prove a new risk bound for this minimax problem in terms of covering numbers under a weak version of Lipschitz condition. Our method can be applied to multi-class classification and popular loss functions including the hinge loss and ramp loss. As some illustrative examples, we derive the adversarial risk bounds for SVMs and deep neural networks, and our bounds have two data-dependent terms, which can be optimized for achieving adversarial robustness.

TIST Journal 2019 Journal Article

Trajectory Data Classification

• Jiang Bian
• Dayong Tian
• Yuanyan Tang
• Dacheng Tao

This article comprehensively surveys the development of trajectory data classification. Considering the critical role of trajectory data classification in modern intelligent systems for surveillance security, abnormal behavior detection, crowd behavior analysis, and traffic control, trajectory data classification has attracted growing attention. According to the availability of manual labels, which is critical to the classification performances, the methods can be classified into three categories, i.e., unsupervised, semi-supervised, and supervised. Furthermore, classification methods are divided into some sub-categories according to what extracted features are used. We provide a holistic understanding and deep insight into three types of trajectory data classification methods and present some promising future directions.

AAAI Conference 2018 Conference Paper

Adversarial Learning of Portable Student Networks

• Yunhe Wang
• Chang Xu
• Chao Xu
• Dacheng Tao

Effective methods for learning deep neural networks with fewer parameters are urgently required, since storage and computations of heavy neural networks have largely prevented their widespread use on mobile devices. Compared with algorithms which directly remove weights or filters for obtaining considerable compression and speed-up ratios, training thin deep networks exploiting the student-teacher learning paradigm is more flexible. However, it is very hard to determine which formulation is optimal to measure the information inherited from teacher networks. To overcome this challenge, we utilize the generative adversarial network (GAN) to learn the student network. In practice, the generator is exactly the student network with extremely less parameters and the discriminator is used as a teaching assistant for distinguishing features extracted from student and teacher networks. By simultaneously optimizing the generator and the discriminator, the resulting student network can produce features of input data with the similar distribution as that of features of the teacher network. Extensive experimental results on benchmark datasets demonstrate that the proposed method is capable of learning well-performed portable networks, which is superior to the state-of-the-art methods.

AAAI Conference 2018 Conference Paper

Domain Generalization via Conditional Invariant Representations

• Ya Li
• Mingming Gong
• Xinmei Tian
• Tongliang Liu
• Dacheng Tao

Domain generalization aims to apply knowledge gained from multiple labeled source domains to unseen target domains. The main difficulty comes from the dataset bias: training data and test data have different distributions, and the training set contains heterogeneous samples from different distributions. Let X denote the features, and Y be the class labels. Existing domain generalization methods address the dataset bias problem by learning a domain-invariant representation h(X) that has the same marginal distribution P(h(X)) across multiple source domains. The functional relationship encoded in P(Y |X) is usually assumed to be stable across domains such that P(Y |h(X)) is also invariant. However, it is unclear whether this assumption holds in practical problems. In this paper, we consider the general situation where both P(X) and P(Y |X) can change across all domains. We propose to learn a feature representation which has domain-invariant class conditional distributions P(h(X)|Y ). With the conditional invariant representation, the invariance of the joint distribution P(h(X), Y ) can be guaranteed if the class prior P(Y ) does not change across training and test domains. Extensive experiments on both synthetic and real data demonstrate the effectiveness of the proposed method.

NeurIPS Conference 2018 Conference Paper

Dual Swap Disentangling

• Zunlei Feng
• Xinchao Wang
• Chenglong Ke
• An-Xiang Zeng
• Dacheng Tao
• Mingli Song

Learning interpretable disentangled representations is a crucial yet challenging task. In this paper, we propose a weakly semi-supervised method, termed as Dual Swap Disentangling (DSD), for disentangling using both labeled and unlabeled data. Unlike conventional weakly supervised methods that rely on full annotations on the group of samples, we require only limited annotations on paired samples that indicate their shared attribute like the color. Our model takes the form of a dual autoencoder structure. To achieve disentangling using the labeled pairs, we follow a encoding-swap-decoding'' process, where we first swap the parts of their encodings corresponding to the shared attribute, and then decode the obtained hybrid codes to reconstruct the original input pairs. For unlabeled pairs, we follow the encoding-swap-decoding'' process twice on designated encoding parts and enforce the final outputs to approximate the input pairs. By isolating parts of the encoding and swapping them back and forth, we impose the dimension-wise modularity and portability of the encodings of the unlabeled samples, which implicitly encourages disentangling under the guidance of labeled pairs. This dual swap mechanism, tailored for semi-supervised setting, turns out to be very effective. Experiments on image datasets from a wide domain show that our model yields state-of-the-art disentangling performances.

NeurIPS Conference 2018 Conference Paper

Learning Versatile Filters for Efficient Convolutional Neural Networks

• Yunhe Wang
• Chang Xu
• Chunjing Xu
• Chao Xu
• Dacheng Tao

This paper introduces versatile filters to construct efficient convolutional neural network. Considering the demands of efficient deep learning techniques running on cost-effective hardware, a number of methods have been developed to learn compact neural networks. Most of these works aim to slim down filters in different ways, e. g. , investigating small, sparse or binarized filters. In contrast, we treat filters from an additive perspective. A series of secondary filters can be derived from a primary filter. These secondary filters all inherit in the primary filter without occupying more storage, but once been unfolded in computation they could significantly enhance the capability of the filter by integrating information extracted from different receptive fields. Besides spatial versatile filters, we additionally investigate versatile filters from the channel perspective. The new techniques are general to upgrade filters in existing CNNs. Experimental results on benchmark datasets and neural networks demonstrate that CNNs constructed with our versatile filters are able to achieve comparable accuracy as that of original filters, but require less memory and FLOPs.

AAAI Conference 2018 Conference Paper

Learning With Single-Teacher Multi-Student

• Shan You
• Chang Xu
• Chao Xu
• Dacheng Tao

In this paper we study a new learning problem defined as “Single-Teacher Multi-Student” (STMS) problem, which investigates how to learn a series of student (simple and specific) models from a single teacher (complex and universal) model. Taking the multiclass and binary classification for example, we focus on learning multiple binary classifiers from a single multiclass classifier, where each of binary classifier is responsible for a certain class. This actually derives from some realistic problems, such as identifying the suspect based on a comprehensive face recognition system. By treating the already-trained multiclass classifier as the teacher, and multiple binary classifiers as the students, we propose a gated support vector machine (gSVM) as a solution. A series of gSVMs are learned with the help of single teacher multiclass classifier. The teacher’s help is two-fold; first, the teacher’s score provides the gated values for students’ decision; second, the teacher can guide the students to accommodate training examples with different difficulty degrees. Extensive experiments on real datasets validate its effectiveness.

IJCAI Conference 2018 Conference Paper

Online Heterogeneous Transfer Metric Learning

• Yong Luo
• Tongliang Liu
• Yonggang Wen
• Dacheng Tao

Distance metric learning (DML) has been demonstrated to be successful and essential in diverse applications. Transfer metric learning (TML) can help DML in the target domain with limited label information by utilizing information from some related source domains. The heterogeneous TML (HTML), where the feature representations vary from the source to the target domain, is general and challenging. However, current HTML approaches are usually conducted in a batch manner and cannot handle sequential data. This motivates the proposed online HTML (OHTML) method. In particular, the distance metric in the source domain is pre-trained using some existing DML algorithms. To enable knowledge transfer, we assume there are large amounts of unlabeled corresponding data that have representations in both the source and target domains. By enforcing the distances (between these unlabeled samples) in the target domain to agree with those in the source domain under the manifold regularization theme, we learn an improved target metric. We formulate the problem in the online setting so that the optimization is efficient and the model can be adapted to new coming data. Experiments in diverse applications demonstrate both effectiveness and efficiency of the proposed method.

IJCAI Conference 2018 Conference Paper

Quantum Divide-and-Conquer Anchoring for Separable Non-negative Matrix Factorization

• Yuxuan Du
• Tongliang Liu
• Yinan Li
• Runyao Duan
• Dacheng Tao

It is NP-complete to find non-negative factors W and H with fixed rank r from a non-negative matrix X by minimizing ||X-WH^Τ ||^2. Although the separability assumption (all data points are in the conical hull of the extreme rows) enables polynomial-time algorithms, the computational cost is not affordable for big data. This paper investigates how the power of quantum computation can be capitalized to solve the non-negative matrix factorization with the separability assumption (SNMF) by devising a quantum algorithm based on the divide-and-conquer anchoring (DCA) scheme [Zhou et al. , 2013]. The design of quantum DCA (QDCA) is challenging. In the divide step, the random projections in DCA is completed by a quantum algorithm for linear operations, which achieves the exponential speedup. We then devise a heuristic post-selection procedure which extracts the information of anchors stored in the quantum states efficiently. Under a plausible assumption, QDCA performs efficiently, achieves the quantum speedup, and is beneficial for high dimensional problems.

IJCAI Conference 2018 Conference Paper

R-SVM+: Robust Learning with Privileged Information

• Xue Li
• Bo Du
• Chang Xu
• Yipeng Zhang
• Lefei Zhang
• Dacheng Tao

In practice, the circumstance that training and test data are clean is not always satisfied. The performance of existing methods in the learning using privileged information (LUPI) paradigm may be seriously challenged, due to the lack of clear strategies to address potential noises in the data. This paper proposes a novel Robust SVM+ (RSVM+) algorithm based on a rigorous theoretical analysis. Under the SVM+ framework in the LUPI paradigm, we study the lower bound of perturbations of both example feature data and privileged feature data, which will mislead the model to make wrong decisions. By maximizing the lower bound, tolerance of the learned model over perturbations will be increased. Accordingly, a novel regularization function is introduced to upgrade a variant form of SVM+. The objective function of RSVM+ is transformed into a quadratic programming problem, which can be efficiently optimized using off-the-shelf solvers. Experiments on real-world datasets demonstrate the necessity of studying robust SVM+ and the effectiveness of the proposed algorithm.

AAAI Conference 2018 Conference Paper

Reinforced Multi-Label Image Classification by Exploring Curriculum

• Shiyi He
• Chang Xu
• Tianyu Guo
• Chao Xu
• Dacheng Tao

Humans and animals learn much better when the examples are not randomly presented but organized in a meaningful order which illustrates gradually more concepts, and gradually more complex ones. Inspired by this curriculum learning mechanism, we propose a reinforced multi-label image classification approach imitating human behavior to label image from easy to complex. This approach allows a reinforcement learning agent to sequentially predict labels by fully exploiting image feature and previously predicted labels. The agent discovers the optimal policies through maximizing the longterm reward which reflects prediction accuracies. Experimental results on PASCAL VOC2007 and 2012 demonstrate the necessity of reinforcement multi-label learning and the algorithm’s effectiveness in real-world multi-label image classification tasks.

IJCAI Conference 2018 Conference Paper

Rethinking Diversified and Discriminative Proposal Generation for Visual Grounding

• Zhou Yu
• Jun Yu
• Chenchao Xiang
• Zhou Zhao
• Qi Tian
• Dacheng Tao

Visual grounding aims to localize an object in an image referred to by a textual query phrase. Various visual grounding approaches have been proposed, and the problem can be modularized into a general framework: proposal generation, multi-modal feature representation, and proposal ranking. Of these three modules, most existing approaches focus on the latter two, with the importance of proposal generation generally neglected. In this paper, we rethink the problem of what properties make a good proposal generator. We introduce the diversity and discrimination simultaneously when generating proposals, and in doing so propose Diversified and Discriminative Proposal Networks model (DDPN). Based on the proposals generated by DDPN, we propose a high performance baseline model for visual grounding and evaluate it on four benchmark datasets. Experimental results demonstrate that our model delivers significant improvements on all the tested data-sets (e. g. , 18. 8% improvement on ReferItGame and 8. 2% improvement on Flickr30k Entities over the existing state-of-the-arts respectively).

IJCAI Conference 2018 Conference Paper

Semantic Structure-based Unsupervised Deep Hashing

• Erkun Yang
• Cheng Deng
• Tongliang Liu
• Wei Liu
• Dacheng Tao

Hashing is becoming increasingly popular for approximate nearest neighbor searching in massive databases due to its storage and search efficiency. Recent supervised hashing methods, which usually construct semantic similarity matrices to guide hash code learning using label information, have shown promising results. However, it is relatively difficult to capture and utilize the semantic relationships between points in unsupervised settings. To address this problem, we propose a novel unsupervised deep framework called Semantic Structure-based unsupervised Deep Hashing (SSDH). We first empirically study the deep feature statistics, and find that the distribution of the cosine distance for point pairs can be estimated by two half Gaussian distributions. Based on this observation, we construct the semantic structure by considering points with distances obviously smaller than the others as semantically similar and points with distances obviously larger than the others as semantically dissimilar. We then design a deep architecture and a pair-wise loss function to preserve this semantic structure in Hamming space. Extensive experiments show that SSDH significantly outperforms current state-of-the-art methods.

AAAI Conference 2018 Conference Paper

Sequence-to-Sequence Learning via Shared Latent Representation

• Xu Shen
• Xinmei Tian
• Jun Xing
• Yong Rui
• Dacheng Tao

Sequence-to-sequence learning is a popular research area in deep learning, such as video captioning and speech recognition. Existing methods model this learning as a mapping process by first encoding the input sequence to a fixed-sized vector, followed by decoding the target sequence from the vector. Although simple and intuitive, such mapping model is task-specific, unable to be directly used for different tasks. In this paper, we propose a star-like framework for general and flexible sequence-to-sequence learning, where different types of media contents (the peripheral nodes) could be encoded to and decoded from a shared latent representation (SLR) (the central node). This is inspired by the fact that human brain could learn and express an abstract concept in different ways. The media-invariant property of SLR could be seen as a high-level regularization on the intermediate vector, enforcing it to not only capture the latent representation intra each individual media like the auto-encoders, but also their transitions like the mapping models. Moreover, the SLR model is content-specific, which means it only needs to be trained once for a dataset, while used for different tasks. We show how to train a SLR model via dropout and use it for different sequence-to-sequence tasks. Our SLR model is validated on the Youtube2Text and MSR-VTT datasets, achieving superior performance on video-to-sentence task, and the first sentence-to-video results.

YNIMG Journal 2018 Journal Article

Variation in longitudinal trajectories of cortical sulci in normal elderly

• Xinke Shen
• Tao Liu
• Dacheng Tao
• Yubo Fan
• Jicong Zhang
• Shuyu Li
• Jiyang Jiang
• Wanlin Zhu

Sulcal morphology has been reported to change with age-related neurological diseases, but the trajectories of sulcal change in normal ageing in the elderly is still unclear. We conducted a study of sulcal morphological changes over seven years in 132 normal elderly participants aged 70–90 years at baseline, and who remained cognitively normal for the next seven years. We examined the fold opening and sulcal depth of sixteen (eight on each hemisphere) prominent sulci based on T1-weighted MRI using automated methods with visual quality control. The trajectory of each individual sulcus with respect to age was examined separately by linear mixed models. Fold opening was best modelled by cubic fits in five sulci, by quadratic models in six sulci and by linear models in five sulci, indicating an accelerated widening of a number of sulci in older age. Sulcal depth showed significant linear decline in three sulci and quadratic trend in one sulcus. Turning points of non-linear trajectories towards accelerated widening of the fold were found to be around the age between 75 and 80, indicating an accelerated atrophy of brain cortex starting in the age of late 70s. Our findings of cortical sulcal changes in normal ageing could provide a reference for studies of neurocognitive disorders, including neurodegenerative diseases, in the elderly.

AAAI Conference 2017 Conference Paper

A Diversified Generative Latent Variable Model for WiFi-SLAM

• Hao Xiong
• Dacheng Tao

WiFi-SLAM aims to map WiFi signals within an unknown environment while simultaneously determining the location of a mobile device. This localization method has been extensively used in indoor, space, undersea, and underground environments. For the sake of accuracy, most methods label the signal readings against ground truth locations. However, this is impractical in large environments, where it is hard to collect and maintain the data. Some methods use latent variable models to generate latent-space locations of signal strength data, an advantage being that no prior labeling of signal strength readings and their physical locations is required. However, the generated latent variables cannot cover all wireless signal locations and WiFi-SLAM performance is significantly degraded. Here we propose the diversified generative latent variable model (DGLVM) to overcome these limitations. By building a positive-definite kernel function, a diversity-encouraging prior is introduced to render the generated latent variables non-overlapping, thus capturing more wireless signal measurements characteristics. The defined objective function is then solved by variational inference. Our experiments illustrate that the method performs WiFi localization more accurately than other label-free methods.

IJCAI Conference 2017 Conference Paper

Adaptive Manifold Regularized Matrix Factorization for Data Clustering

• Lefei Zhang
• Qian Zhang
• Bo Du
• Jane You
• Dacheng Tao

Data clustering is the task to group the data samples into certain clusters based on the relationships of samples and structures hidden in data, and it is a fundamental and important topic in data mining and machine learning areas. In the literature, the spectral clustering is one of the most popular approaches and has many variants in recent years. However, the performance of spectral clustering is determined by the affinity matrix, which is always computed by a predefined model (e. g. , Gaussian kernel function) with carefully tuned parameters combination, and may far from optimal in practice. In this paper, we propose to consider the observed data clustering as a robust matrix factorization point of view, and learn an affinity matrix simultaneously to regularize the proposed matrix factorization. The solution of the proposed adaptive manifold regularized matrix factorization (AMRMF) is reached by a novel Augmented Lagrangian Multiplier (ALM) based algorithm. The experimental results on standard clustering datasets demonstrate the superior performance over the exist alternatives.

ICML Conference 2017 Conference Paper

Algorithmic Stability and Hypothesis Complexity

• Tongliang Liu
• Gábor Lugosi
• Gergely Neu
• Dacheng Tao

We introduce a notion of algorithmic stability of learning algorithms—that we term hypothesis stability —that captures stability of the hypothesis output by the learning algorithm in the normed space of functions from which hypotheses are selected. The main result of the paper bounds the generalization error of any learning algorithm in terms of its hypothesis stability. The bounds are based on martingale inequalities in the Banach space to which the hypotheses belong. We apply the general bounds to bound the performance of some learning algorithms based on empirical risk minimization and stochastic gradient descent.

ICML Conference 2017 Conference Paper

Beyond Filters: Compact Feature Map for Portable Deep Model

• Yunhe Wang 0001
• Chang Xu 0002
• Chao Xu 0006
• Dacheng Tao

Convolutional neural networks (CNNs) have shown extraordinary performance in a number of applications, but they are usually of heavy design for the accuracy reason. Beyond compressing the filters in CNNs, this paper focuses on the redundancy in the feature maps derived from the large number of filters in a layer. We propose to extract intrinsic representation of the feature maps and preserve the discriminability of the features. Circulant matrix is employed to formulate the feature map transformation, which only requires O(dlog d) computation complexity to embed a d-dimensional feature map. The filter is then re-configured to establish the mapping from original input to the new compact feature map, and the resulting network can preserve intrinsic information of the original network with significantly fewer parameters, which not only decreases the online memory for launching CNN but also accelerates the computation speed. Experiments on benchmark image datasets demonstrate the superiority of the proposed algorithm over state-of-the-art methods.

AAAI Conference 2017 Conference Paper

Beyond RPCA: Flattening Complex Noise in the Frequency Domain

• Yunhe Wang
• Chang Xu
• Chao Xu
• Dacheng Tao

Discovering robust low-rank data representations is important in many real-world problems. Traditional robust principal component analysis (RPCA) assumes that the observed data are corrupted by some sparse noise (e. g. , Laplacian noise) and utilizes the 1-norm to separate out the noisy component. Nevertheless, as well as simple Gaussian or Laplacian noise, noise in real-world data is often more complex, and thus the 1 and 2-norms are insufficient for noise characterization. This paper presents a more flexible approach to modeling complex noise by investigating their properties in the frequency domain. Although elements of a noise matrix are chaotic in the spatial domain, the absolute values of its alternative coefficients in the frequency domain are constant w. r. t. their variance. Based on this observation, a new robust PCA algorithm is formulated by simultaneously discovering the low-rank and noisy components. Extensive experiments on synthetic data and video background subtraction demonstrate that FRPCA is effective for handles complex noise.

UAI Conference 2017 Conference Paper

Causal Discovery from Temporally Aggregated Time Series

• Mingming Gong
• Kun Zhang 0001
• Bernhard Schölkopf
• Clark Glymour
• Dacheng Tao

Discovering causal structure of a dynamical system from observed time series is a traditional and important problem. In many practical applications, observed data are obtained by applying subsampling or temporally aggregation to the original causal processes, making it difficult to discover the underlying causal relations. Subsampling refers to the procedure that for every k consecutive observations, one is kept, the rest being skipped, and recently some advances have been made in causal discovery from such data. With temporal aggregation, the local averages or sums of k consecutive, non-overlapping observations in the causal process are computed as new observations, and causal discovery from such data is even harder. In this paper, we investigate how to recover causal relations at the original causal frequency from temporally aggregated data when k is known. Assuming the time series at the causal frequency follows a vector autoregressive (VAR) model, we show that the causal structure at the causal frequency is identifiable from aggregated time series if the noise terms are independent and non-Gaussian and some other technical conditions hold. We then present an estimation method based on non-Gaussian state-space modeling and evaluate its performance on both synthetic and real data.

IJCAI Conference 2017 Conference Paper

Classification and Representation Joint Learning via Deep Networks

• Ya Li
• Xinmei Tian
• Xu Shen
• Dacheng Tao

Deep learning has been proven to be effective for classification problems. However, the majority of previous works trained classifiers by considering only class label information and ignoring the local information from the spatial distribution of training samples. In this paper, we propose a deep learning framework that considers both class label information and local spatial distribution information between training samples. A two-channel network with shared weights is used to measure the local distribution. The classification performance can be improved with more detailed information provided by the local distribution, particularly when the training samples are insufficient. Additionally, the class label information can help to learn better feature representations compared with other feature learning methods that use only local distribution information between samples. The local distribution constraint between sample pairs can also be viewed as a regularization of the network, which can efficiently prevent the overfitting problem. Extensive experiments are conducted on several benchmark image classification datasets, and the results demonstrate the effectiveness of our proposed method.

IJCAI Conference 2017 Conference Paper

Collaborative Rating Allocation

• Yali Du
• Chang Xu
• Dacheng Tao

This paper studies the collaborative rating allocation problem, in which each user has limited ratings on all items. These users are termed ``energy limited''. Different from existing methods which treat each rating independently, we investigate the geometric properties of a user's rating vector, and design a matrix completion method on the simplex. In this method, a user's rating vector is estimated by the combination of user profiles as basis points on the simplex. Instead of using Euclidean metric, a non-linear pull-back distance measurement from the sphere is adopted since it can depict the geometric constraints on each user's rating vector. The resulting objective function is then efficiently optimized by a Riemannian conjugate gradient method on the simplex. Experiments on real-world data sets demonstrate our model's competitiveness versus other collaborative rating prediction methods.

AAAI Conference 2017 Conference Paper

Cost-Sensitive Feature Selection via F-Measure Optimization Reduction

• Meng Liu
• Chang Xu
• Yong Luo
• Chao Xu
• Yonggang Wen
• Dacheng Tao

Feature selection aims to select a small subset from the high-dimensional features which can lead to better learning performance, lower computational complexity, and better model readability. The class imbalance problem has been neglected by traditional feature selection methods, therefore the selected features will be biased towards the majority classes. Because of the superiority of F-measure to accuracy for imbalanced data, we propose to use F-measure as the performance measure for feature selection algorithms. As a pseudo-linear function, the optimization of F-measure can be achieved by minimizing the total costs. In this paper, we present a novel cost-sensitive feature selection (CSFS) method which optimizes F-measure instead of accuracy to take class imbalance issue into account. The features will be selected according to optimal F-measure classifier after solving a series of cost-sensitive feature selection sub-problems. The features selected by our method will fully represent the characteristics of not only majority classes, but also minority classes. Extensive experimental results conducted on synthetic, multi-class and multi-label datasets validate the efficiency and significance of our feature selection method.

IJCAI Conference 2017 Conference Paper

Exploiting High-Order Information in Heterogeneous Multi-Task Feature Learning

• Yong Luo
• Dacheng Tao
• Yonggang Wen

Multi-task feature learning (MTFL) aims to improve the generalization performance of multiple related learning tasks by sharing features between them. It has been successfully applied to many pattern recognition and biometric prediction problems. Most of current MTFL methods assume that different tasks exploit the same feature representation, and thus are not applicable to the scenarios where data are drawn from heterogeneous domains. Existing heterogeneous transfer learning (including multi-task learning) approaches handle multiple heterogeneous domains by usually learning feature transformations across different domains, but they ignore the high-order statistics (correlation information) which can only be discovered by simultaneously exploring all domains. We therefore develop a tensor based heterogeneous MTFL (THMTFL) framework to exploit such high-order information. Specifically, feature transformations of all domains are learned together, and finally used to derive new representations. A connection between all domains is built by using the transformations to project the pre-learned predictive structures of different domains into a common subspace, and minimizing their divergence in the subspace. By exploring the high-order information, the proposed THMTFL can obtain more reliable feature transformations compared with existing heterogeneous transfer learning approaches. Extensive experiments on both text categorization and social image annotation demonstrate superiority of the proposed method.

IJCAI Conference 2017 Conference Paper

Fast SVM Trained by Divide-and-Conquer Anchors

• Meng Liu
• Chang Xu
• Chao Xu
• Dacheng Tao

Supporting vector machine (SVM) is the most frequently used classifier for machine learning tasks. However, its training time could become cumbersome when the size of training data is very large. Thus, many kinds of representative subsets are chosen from the original dataset to reduce the training complexity. In this paper, we propose to choose the representative points which are noted as anchors obtained from non-negative matrix factorization (NMF) in a divide-and-conquer framework, and then use the anchors to train an approximate SVM. Our theoretical analysis shows that the solving the DCA-SVM can yield an approximate solution close to the primal SVM. Experimental results on multiple datasets demonstrate that our DCA-SVM is faster than the state-of-the-art algorithms without notably decreasing the accuracy of classification results.

IJCAI Conference 2017 Conference Paper

General Heterogeneous Transfer Distance Metric Learning via Knowledge Fragments Transfer

• Yong Luo
• Yonggang Wen
• Tongliang Liu
• Dacheng Tao

Transfer learning aims to improve the performance of target learning task by leveraging information (or transferring knowledge) from other related tasks. Recently, transfer distance metric learning (TDML) has attracted lots of interests, but most of these methods assume that feature representations for the source and target learning tasks are the same. Hence, they are not suitable for the applications, in which the data are from heterogeneous domains (feature spaces, modalities and even semantics). Although some existing heterogeneous transfer learning (HTL) approaches is able to handle such domains, they lack flexibility in real-world applications, and the learned transformations are often restricted to be linear. We therefore develop a general and flexible heterogeneous TDML (HTDML) framework based on the knowledge fragment transfer strategy. In the proposed HTDML, any (linear or nonlinear) distance metric learning algorithms can be employed to learn the source metric beforehand. Then a set of knowledge fragments are extracted from the pre-learned source metric to help target metric learning. In addition, either linear or nonlinear distance metric can be learned for the target domain. Extensive experiments on both scene classification and object recognition demonstrate superiority of the proposed method.

IJCAI Conference 2017 Conference Paper

Improving Stochastic Block Models by Incorporating Power-Law Degree Characteristic

• Maoying Qiao
• Jun Yu
• Wei Bian
• Qiang Li
• Dacheng Tao

Stochastic block models (SBMs) provide a statistical way modeling network data, especially in representing clusters or community structures. However, most block models do not consider complex characteristics of networks such as scale-free feature, making them incapable of handling degree variation of vertices, which is ubiquitous in real networks. To address this issue, we introduce degree decay variables into SBM, termed power-law degree SBM (PLD-SBM), to model the varying probability of connections between node pairs. The scale-free feature is approximated by a power-law degree characteristic. Such a property allows PLD-SBM to correct the distortion of degree distribution in SBM, and thus improves the performance of cluster prediction. Experiments on both simulated networks and two real-world networks including the Adolescent Health Data and the political blogs network demonstrate the validity of the motivation of PLD-SBM, and its practical superiority.

TIST Journal 2017 Journal Article

Large Sparse Cone Non-negative Matrix Factorization for Image Annotation

• Dapeng Tao
• Dacheng Tao
• Xuelong Li
• Xinbo Gao

Image annotation assigns relevant tags to query images based on their semantic contents. Since Non-negative Matrix Factorization (NMF) has the strong ability to learn parts-based representations, recently, a number of algorithms based on NMF have been proposed for image annotation and have achieved good performance. However, most of the efforts have focused on the representations of images and annotations. The properties of the semantic parts have not been well studied. In this article, we revisit the sparseness-constrained NMF (sNMF) proposed by Hoyer [2004]. By endowing the sparseness constraint with a geometric interpretation and sNMF with theoretical analyses of the generalization ability, we show that NMF with such a sparseness constraint has three advantages for image annotation tasks: (i) The sparseness constraint is more ℓ 0 -norm oriented than the ℓ 1 -norm-based sparseness, which significantly enhances the ability of NMF to robustly learn semantic parts. (ii) The sparseness constraint has a large cone interpretation and thus allows the reconstruction error of NMF to be smaller, which means that the learned semantic parts are more powerful to represent images for tagging. (iii) The learned semantic parts are less correlated, which increases the discriminative ability for annotating images. Moreover, we present a new efficient large sparse cone NMF (LsCNMF) algorithm to optimize the sNMF problem by employing the Nesterov’s optimal gradient method. We conducted experiments on the PASCAL VOC07 dataset and demonstrated the effectiveness of LsCNMF for image annotation.

IJCAI Conference 2017 Conference Paper

Multi-Positive and Unlabeled Learning

• Yixing Xu
• Chang Xu
• Chao Xu
• Dacheng Tao

The positive and unlabeled (PU) learning problem focuses on learning a classifier from positive and unlabeled data. Some methods have been developed to solve the PU learning problem. However, they are often limited in practical applications, since only binary classes are involved and cannot easily be adapted to multi-class data. Here we propose a one-step method that directly enables multi-class model to be trained using the given input multi-class data and that predicts the label based on the model decision. Specifically, we construct different convex loss functions for labeled and unlabeled data to learn a discriminant function F. The theoretical analysis on the generalization error bound shows that it is no worse than k√k times of the fully supervised multi-class classification methods when the size of the data in k classes is of the same order. Finally, our experimental results demonstrate the significance and effectiveness of the proposed algorithm in synthetic and real-world datasets.

TIST Journal 2017 Journal Article

Multifeature Anisotropic Orthogonal Gaussian Process for Automatic Age Estimation

• Zhifeng Li
• Dihong Gong
• Kai Zhu
• Dacheng Tao
• Xuelong Li

Automatic age estimation is an important yet challenging problem. It has many promising applications in social media. Of the existing age estimation algorithms, the personalized approaches are among the most popular ones. However, most person-specific approaches rely heavily on the availability of training images across different ages for a single subject, which is usually difficult to satisfy in practical application of age estimation. To address this limitation, we first propose a new model called Orthogonal Gaussian Process (OGP), which is not restricted by the number of training samples per person. In addition, without sacrifice of discriminative power, OGP is much more computationally efficient than the standard Gaussian Process. Based on OGP, we then develop an effective age estimation approach, namely anisotropic OGP (A-OGP), to further reduce the estimation error. A-OGP is based on an anisotropic noise level learning scheme that contributes to better age estimation performance. To finally optimize the performance of age estimation, we propose a multifeature A-OGP fusion framework that uses multiple features combined with a random sampling method in the feature space. Extensive experiments on several public domain face aging datasets (FG-NET, MORPH Album1, and MORPH Album 2) are conducted to demonstrate the state-of-the-art estimation accuracy of our new algorithms.

IJCAI Conference 2017 Conference Paper

On Gleaning Knowledge from Multiple Domains for Active Learning

• Zengmao Wang
• Bo Du
• Lefei Zhang
• Liangpei Zhang
• Ruimin Hu
• Dacheng Tao

How can a doctor diagnose new diseases with little historical knowledge, which are emerging over time? Active learning is a promising way to address the problem by querying the most informative samples. Since the diagnosed cases for new disease are very limited, gleaning knowledge from other domains (classical prescriptions) to prevent the bias of active leaning would be vital for accurate diagnosis. In this paper, a framework that attempts to glean knowledge from multiple domains for active learning by querying the most uncertain and representative samples from the target domain and calculating the importance weights for re-weighting the source data in a single unified formulation is proposed. The weights are optimized by both a supervised classifier and distribution matching between the source domain and target domain with maximum mean discrepancy. Besides, a multiple domains active learning method is designed based on the proposed framework as an example. The proposed method is verified with newsgroups and handwritten digits data recognition tasks, where it outperforms the state-of-the-art methods.

AAAI Conference 2017 Conference Paper

Pairwise Relationship Guided Deep Hashing for Cross-Modal Retrieval

• Erkun Yang
• Cheng Deng
• Wei Liu
• Xianglong Liu
• Dacheng Tao
• Xinbo Gao

With benefits of low storage cost and fast query speed, crossmodal hashing has received considerable attention recently. However, almost all existing methods on cross-modal hashing cannot obtain powerful hash codes due to directly utilizing hand-crafted features or ignoring heterogeneous correlations across different modalities, which will greatly degrade the retrieval performance. In this paper, we propose a novel deep cross-modal hashing method to generate compact hash codes through an end-to-end deep learning architecture, which can effectively capture the intrinsic relationships between various modalities. Our architecture integrates different types of pairwise constraints to encourage the similarities of the hash codes from an intra-modal view and an inter-modal view, respectively. Moreover, additional decorrelation constraints are introduced to this architecture, thus enhancing the discriminative ability of each hash bit. Extensive experiments show that our proposed method yields state-of-the-art results on two cross-modal retrieval datasets.

AAAI Conference 2017 Conference Paper

Patch Reordering: A NovelWay to Achieve Rotation and Translation Invariance in Convolutional Neural Networks

• Xu Shen
• Xinmei Tian
• Shaoyan Sun
• Dacheng Tao

Convolutional Neural Networks (CNNs) have demonstrated state-of-the-art performance on many visual recognition tasks. However, the combination of convolution and pooling operations only shows invariance to small local location changes in meaningful objects in input. Sometimes, such networks are trained using data augmentation to encode this invariance into the parameters, which restricts the capacity of the model to learn the content of these objects. A more ef- ficient use of the parameter budget is to encode rotation or translation invariance into the model architecture, which relieves the model from the need to learn them. To enable the model to focus on learning the content of objects other than their locations, we propose to conduct patch ranking of the feature maps before feeding them into the next layer. When patch ranking is combined with convolution and pooling operations, we obtain consistent representations despite the location of meaningful objects in input. We show that the patch ranking module improves the performance of the CNN on many benchmark tasks, including MNIST digit recognition, large-scale image recognition, and image retrieval.

IJCAI Conference 2017 Conference Paper

Privileged Matrix Factorization for Collaborative Filtering

• Yali Du
• Chang Xu
• Dacheng Tao

Collaborative filtering plays a crucial role in reducing excessive information in online consuming by suggesting products to customers that fulfil their potential interests. Observing that a user's review comments on purchases are often in companion with ratings, recent works exploit the review texts in representing user or item factors and have achieved prominent performance. Although effectiveness of reviews has been verified, one major defect of existing works is that reviews are used in justifying the learning of either user or item factors without noticing that each review associates a pair of user and item concurrently. To better explore the value of review comments, this paper presents the privileged matrix factorization method that utilize reviews in the learning of both user and item factors. By mapping review texts into the privileged feature space, a learned privileged function compensates the discrepancies between predicted ratings and groundtruth values rating-wisely. Thus by minimizing discrepancies and prediction errors, our method harnesses the information present in the review comments for the learning of both user and item factors. Experiments on five real datasets testify the effectiveness of the proposed method.

IJCAI Conference 2017 Conference Paper

Privileged Multi-label Learning

• Shan You
• Chang Xu
• Yunhe Wang
• Chao Xu
• Dacheng Tao

This paper presents privileged multi-label learning (PrML) to explore and exploit the relationship between labels in multi-label learning problems. We suggest that for each individual label, it cannot only be implicitly connected with other labels via the low-rank constraint over label predictors, but also its performance on examples can receive the explicit comments from other labels together acting as an Oracle teacher. We generate privileged label feature for each example and its individual label, and then integrate it into the framework of low-rank based multi-label learning. The proposed algorithm can therefore comprehensively explore and exploit label relationships by inheriting all the merits of privileged information and low-rank constraints. We show that PrML can be efficiently solved by dual coordinate descent algorithm using iterative optimization strategy with cheap updates. Experiments on benchmark datasets show that through privileged label features, the performance can be significantly improved and PrML is superior to several competing methods in most cases.

AAAI Conference 2017 Conference Paper

Robust Manifold Matrix Factorization for Joint Clustering and Feature Extraction

• Lefei Zhang
• Qian Zhang
• Bo Du
• Dacheng Tao
• Jane You

Low-rank matrix approximation has been widely used for data subspace clustering and feature representation in many computer vision and pattern recognition applications. However, in order to enhance the discriminability, most of the matrix approximation based feature extraction algorithms usually generate the cluster labels by certain clustering algorithm (e. g. , the kmeans) and then perform the matrix approximation guided by such label information. In addition, the noises and outliers in the dataset with large reconstruction errors will easily dominate the objective function by the conventional 2norm based squared residue minimization. In this paper, we propose a novel clustering and feature extraction algorithm based on an unified low-rank matrix factorization framework, which suggests that the observed data matrix can be approximated by the production of projection matrix and low dimensional representation, among which the low-dimensional representation can be approximated by the cluster indicator and latent feature matrix simultaneously. Furthermore, we have proposed using the 2, 1-norm and integrating the manifold regularization to further promote the proposed model. A novel Augmented Lagrangian Method (ALM) based procedure is designed to effectively and efficiently seek the optimal solution of the problem. The experimental results in both clustering and feature extraction perspectives demonstrate the superior performance of the proposed method.

AAAI Conference 2017 Conference Paper

Structure Regularized Unsupervised Discriminant Feature Analysis

• Mingyu Fan
• Xiaojun Chang
• Dacheng Tao

Feature selection is an important technique in machine learning research. An effective and robust feature selection method is desired to simultaneously identify the informative features and eliminate the noisy ones of data. In this paper, we consider the unsupervised feature selection problem which is particularly difficult as there is not any class labels that would guide the search for relevant features. To solve this, we propose a novel algorithmic framework which performs unsupervised feature selection. Firstly, the proposed framework implements structure learning, where the data structures (including intrinsic distribution structure and the data segment) are found via a combination of the alternative optimization and clustering. Then, both the intrinsic data structure and data segmentation are formulated as regularization terms for discriminant feature selection. The results of the feature selection also affect the structure learning step in the following iterations. By leveraging the interactions between structure learning and feature selection, we are able to capture more accurate structure of data and select more informative features. Clustering and classification experiments on real world image data sets demonstrate the effectiveness of our method.

IJCAI Conference 2017 Conference Paper

Tag Disentangled Generative Adversarial Network for Object Image Re-rendering

• Chaoyue Wang
• Chaohui Wang
• Chang Xu
• Dacheng Tao

In this paper, we propose a principled Tag Disentangled Generative Adversarial Networks (TD-GAN) for re-rendering new images for the object of interest from a single image of it by specifying multiple scene properties (such as viewpoint, illumination, expression, etc. ). The whole framework consists of a disentangling network, a generative network, a tag mapping net, and a discriminative network, which are trained jointly based on a given set of images that are completely/partially tagged (i. e. , supervised/semi-supervised setting). Given an input image, the disentangling network extracts disentangled and interpretable representations, which are then used to generate images by the generative network. In order to boost the quality of disentangled representations, the tag mapping net is integrated to explore the consistency between the image and its tags. Furthermore, the discriminative network is introduced to implement the adversarial training strategy for generating more realistic images. Experiments on two challenging datasets demonstrate the state-of-the-art performance of the proposed framework in the problem of interest.

IJCAI Conference 2017 Conference Paper

Understanding How Feature Structure Transfers in Transfer Learning

• Tongliang Liu
• Qiang Yang
• Dacheng Tao

Transfer learning transfers knowledge across domains to improve the learning performance. Since feature structures generally represent the common knowledge across different domains, they can be transferred successfully even though the labeling functions across domains differ arbitrarily. However, theoretical justification for this success has remained elusive. In this paper, motivated by self-taught learning, we regard a set of bases as a feature structure of a domain if the bases can (approximately) reconstruct any observation in this domain. We propose a general analysis scheme to theoretically justify that if the source and target domains share similar feature structures, the source domain feature structure is transferable to the target domain, regardless of the change of the labeling functions across domains. The transferred structure is interpreted to function as a regularization matrix which benefits the learning process of the target domain task. We prove that such transfer enables the corresponding learning algorithms to be uniformly stable. Specifically, we illustrate the existence of feature structure transfer in two well-known transfer learning settings: domain adaptation and learning to learn.

TIST Journal 2016 Journal Article

A Comprehensive Survey on Pose-Invariant Face Recognition

• Changxing Ding
• Dacheng Tao

The capacity to recognize faces under varied poses is a fundamental human ability that presents a unique challenge for computer vision systems. Compared to frontal face recognition, which has been intensively studied and has gradually matured in the past few decades, Pose-Invariant Face Recognition (PIFR) remains a largely unsolved problem. However, PIFR is crucial to realizing the full potential of face recognition for real-world applications, since face recognition is intrinsically a passive biometric technology for recognizing uncooperative subjects. In this article, we discuss the inherent difficulties in PIFR and present a comprehensive review of established techniques. Existing PIFR methods can be grouped into four categories, that is, pose-robust feature extraction approaches, multiview subspace learning approaches, face synthesis approaches, and hybrid approaches. The motivations, strategies, pros/cons, and performance of representative approaches are described and compared. Moreover, promising directions for future research are discussed.

NeurIPS Conference 2016 Conference Paper

CNNpack: Packing Convolutional Neural Networks in the Frequency Domain

• Yunhe Wang
• Chang Xu
• Shan You
• Dacheng Tao
• Chao Xu

Deep convolutional neural networks (CNNs) are successfully used in a number of applications. However, their storage and computational requirements have largely prevented their widespread use on mobile devices. Here we present an effective CNN compression approach in the frequency domain, which focuses not only on smaller weights but on all the weights and their underlying connections. By treating convolutional filters as images, we decompose their representations in the frequency domain as common parts (i. e. , cluster centers) shared by other similar filters and their individual private parts (i. e. , individual residuals). A large number of low-energy frequency coefficients in both parts can be discarded to produce high compression without significantly compromising accuracy. We relax the computational burden of convolution operations in CNNs by linearly combining the convolution responses of discrete cosine transform (DCT) bases. The compression and speed-up ratios of the proposed algorithm are thoroughly analyzed and evaluated on benchmark image datasets to demonstrate its superiority over state-of-the-art methods.

AAAI Conference 2016 Conference Paper

Discriminative Vanishing Component Analysis

• Chenping Hou
• Feiping Nie
• Dacheng Tao

Vanishing Component Analysis (VCA) is a recently proposed prominent work in machine learning. It narrows the gap between tools and computational algebra: the vanishing ideal and its applications to classification problem. In this paper, we will analyze VCA in the kernel view, which is also another important research direction in machine learning. Under a very weak assumption, we provide a different point of view to VCA and make the kernel trick on VCA become possible. We demonstrate that the projection matrix derived by VCA is located in the same space as that of Kernel Principal Component Analysis (KPCA) with a polynomial kernel. Two groups of projections can express each other by linear transformation. Furthermore, we prove that KPCA and VCA have identical discriminative power, provided that the ratio trace criteria is employed as the measurement. We also show that the kernel formulated by the inner products of VCA’s projections can be expressed by the KPCA’s kernel linearly. Based on the analysis above, we proposed a novel Discriminative Vanishing Component Analysis (DVCA) approach. Experimental results are provided for demonstration.

AAAI Conference 2016 Conference Paper

Diversified Dynamical Gaussian Process Latent Variable Model for Video Repair

• Hao Xiong
• Tongliang Liu
• Dacheng Tao

Videos can be conserved on different media. However, storing on media such as films and hard disks can suffer from unexpected data loss, for instance from physical damage. Repair of missing or damaged pixels is essential for video maintenance and preservation. Most methods seek to fill in missing holes by synthesizing similar textures from local or global frames. However, this can introduce incorrect contexts, especially when the missing hole or number of damaged frames is large. Furthermore, simple texture synthesis can introduce artifacts in undamaged and recovered areas. To address aforementioned problems, we propose the diversified dynamical Gaussian process latent variable model (D2 GPLVM) for considering the variety in existing videos and thus introducing a diversity encouraging prior to inducing points. The aim is to ensure that the trained inducing points, which are a smaller set of all observed undamaged frames, are more diverse and resistant for contextaware and artifacts-free based video repair. The defined objective function in our proposed model is initially not analytically tractable and must be solved by variational inference. Finally, experimental testing illustrates the robustness and effectiveness of our method for damaged video repair.

ICML Conference 2016 Conference Paper

Domain Adaptation with Conditional Transferable Components

• Mingming Gong
• Kun Zhang 0001
• Tongliang Liu
• Dacheng Tao
• Clark Glymour
• Bernhard Schölkopf

Domain adaptation arises in supervised learning when the training (source domain) and test (target domain) data have different distributions. Let X and Y denote the features and target, respectively, previous work on domain adaptation considers the covariate shift situation where the distribution of the features P(X) changes across domains while the conditional distribution P(Y|X) stays the same. To reduce domain discrepancy, recent methods try to find invariant components \mathcalT(X) that have similar P(\mathcalT(X)) by explicitly minimizing a distribution discrepancy measure. However, it is not clear if P(Y|\mathcalT(X)) in different domains is also similar when P(Y|X) changes. Furthermore, transferable components do not necessarily have to be invariant. If the change in some components is identifiable, we can make use of such components for prediction in the target domain. In this paper, we focus on the case where P(X|Y) and P(Y) both change in a causal system in which Y is the cause for X. Under appropriate assumptions, we aim to extract conditional transferable components whose conditional distribution P(\mathcalT(X)|Y) is invariant after proper location-scale (LS) transformations, and identify how P(Y) changes between domains simultaneously. We provide theoretical analysis and empirical evaluation on both synthetic and real-world data to show the effectiveness of our method.

IJCAI Conference 2016 Conference Paper

Empirical Risk Minimization for Metric Learning Using Privileged Information

• Xun Yang
• Meng Wang
• Luming Zhang
• Dacheng Tao

Traditional metric learning methods usually make decisions based on a fixed threshold, which may result in a suboptimal metric when the inter-class and inner-class variations are complex. To address this issue, in this paper we propose an effective metric learning method by exploiting privileged information to relax the fixed threshold under the empirical risk minimization framework. Privileged information describes useful high-level semantic information that is only available during training. Our goal is to improve the performance by incorporating privileged information to design a locally adaptive decision function. We jointly learn two distance metrics by minimizing the empirical loss penalizing the difference between the distance in the original space and that in the privileged space. The distance in the privileged space functions as a locally adaptive decision threshold, which can guide the decision making like a teacher. We optimize the objective function using the Accelerated Proximal Gradient approach to obtain a global optimum solution. Experiment results show that by leveraging privileged information, our proposed method can achieve satisfactory performance.

TIST Journal 2016 Journal Article

Joint Structured Sparsity Regularized Multiview Dimension Reduction for Video-Based Facial Expression Recognition

• Liping Xie
• Dacheng Tao
• Haikun Wei

Video-based facial expression recognition (FER) has recently received increased attention as a result of its widespread application. Using only one type of feature to describe facial expression in video sequences is often inadequate, because the information available is very complex. With the emergence of different features to represent different properties of facial expressions in videos, an appropriate combination of these features becomes an important, yet challenging, problem. Considering that the dimensionality of these features is usually high, we thus introduce multiview dimension reduction (MVDR) into video-based FER. In MVDR, it is critical to explore the relationships between and within different feature views. To achieve this goal, we propose a novel framework of MVDR by enforcing joint structured sparsity at both inter- and intraview levels. In this way, correlations on and between the feature spaces of different views tend to be well-exploited. In addition, a transformation matrix is learned for each view to discover the patterns contained in the original features, so that the different views are comparable in finding a common representation. The model can be not only performed in an unsupervised manner, but also easily extended to a semisupervised setting by incorporating some domain knowledge. An alternating algorithm is developed for problem optimization, and each subproblem can be efficiently solved. Experiments on two challenging video-based FER datasets demonstrate the effectiveness of the proposed framework.

AAAI Conference 2016 Conference Paper

Linear Submodular Bandits with a Knapsack Constraint

• Baosheng Yu
• Meng Fang
• Dacheng Tao

Linear submodular bandits has been proven to be effective in solving the diversification and feature-based exploration problems in retrieval systems. Concurrently, many web-based applications, such as news article recommendation and online ad placement, can be modeled as budget-limited problems. However, the diversification problem under a budget constraint has not been considered. In this paper, we first introduce the budget constraint to linear submodular bandits as a new problem called the linear submodular bandits with a knapsack constraint. We then define an α-approximation unit-cost regret considering that submodular function maximization is NP-hard. To solve this problem, we propose two greedy algorithms based on a modified UCB rule. We then prove these two algorithms with different regret bounds and computational costs. We also conduct a number of experiments and the experimental results confirm our theoretical analyses.

IJCAI Conference 2016 Conference Paper

Multi-View Exclusive Unsupervised Dimension Reduction for Video-Based Facial Expression Recognition

• Liping Xie
• Dacheng Tao
• Haikun Wei

Video-based facial expression recognition (FER)has recently received increased attention as a result of its widespread application. Many kinds of features have been proposed to represent different properties of facial expressions in videos. However the dimensionality of these features is usually high. In addition, due to the complexity of the information available in video sequences, using only one type of feature is often inadequate. How to effectively reduce the dimensionality and combine multi-view features thus becomes a challenging problem. In this paper, motivated by the recent success in exclusive feature selection, we first introduce exclusive group LASSO (EG-LASSO) to unsupervised dimension reduction (UDR). This leads to the proposed exclusive UDR (EUDR) framework, which allows arbitrary sparse structures on the feature space. To properly combine multiple kinds of features, we further extend EUDR to multi-view EUDR (MEUDR), where the structured sparsity is enforced at both intra- and inter-view levels. In addition, combination weights are learned for all views to allow them to contribute differently to the final consensus presentation. A reliable solution is then obtained. Experiments on two challenging video-based FER datasets demonstrate the effectiveness of the proposed method.

TIST Journal 2016 Journal Article

Mutual Component Analysis for Heterogeneous Face Recognition

• Zhifeng Li
• Dihong Gong
• Qiang Li
• Dacheng Tao
• Xuelong Li

Heterogeneous face recognition, also known as cross-modality face recognition or intermodality face recognition, refers to matching two face images from alternative image modalities. Since face images from different image modalities of the same person are associated with the same face object, there should be mutual components that reflect those intrinsic face characteristics that are invariant to the image modalities. Motivated by this rationality, we propose a novel approach called Mutual Component Analysis (MCA) to infer the mutual components for robust heterogeneous face recognition. In the MCA approach, a generative model is first proposed to model the process of generating face images in different modalities, and then an Expectation Maximization (EM) algorithm is designed to iteratively learn the model parameters. The learned generative model is able to infer the mutual components (which we call the hidden factor, where hidden means the factor is unreachable and invisible, and can only be inferred from observations) that are associated with the person’s identity, thus enabling fast and effective matching for cross-modality face recognition. To enhance recognition performance, we propose an MCA-based multiclassifier framework using multiple local features. Experimental results show that our new approach significantly outperforms the state-of-the-art results on two typical application scenarios: sketch-to-photo and infrared-to-visible face recognition.

IJCAI Conference 2016 Conference Paper

On Combining Side Information and Unlabeled Data for Heterogeneous Multi-Task Metric Learning

• Yong Luo
• Yonggang Wen
• Dacheng Tao

Distance metric learning (DML) is critical for a wide variety of machine learning algorithms and pattern recognition applications. Transfer metric learning (TML) leverages the side information (e. g. , similar/dissimilar constraints over pairs of samples) from related domains to help the target metric learning (with limited information). Current TML tools usually assume that different domains exploit the same feature representation, and thus are not applicable to tasks where data are drawn from heterogeneous domains. Heterogeneous transfer learning approaches handle heterogeneous domains by usually learning feature transformations across different domains. The learned transformation can be used to derive a metric, but these approaches are mostly limited by their capability of only handling two domains. This motivates the proposed heterogeneous multi-task metric learning (HMTML) framework for handling multiple domains by combining side information and unlabeled data. Specifically, HMTML learns the metrics for all different domains simultaneously by maximizing their high-order correlation (parameterized by feature covariance of unlabeled data) in a common subspace, which is induced by the transformations derived from the metrics. Extensive experiments on both multi-language text categorization and multi-view social image annotation demonstrate the effectiveness of the proposed method.

IJCAI Conference 2016 Conference Paper

Online Multi-Object Tracking by Quadratic Pseudo-Boolean Optimization

• Long Lan
• Dacheng Tao
• Chen Gong
• Naiyang Guan
• Zhigang Luo

Online multi-object tracking (MOT) is challenging: frame-by-frame matching of detection hypotheses to the correct trackers can be difficult. The Hungarian algorithm is the most commonly used online MOT data association method due to its rapid assignment; however, the Hungarian algorithm simply considers associations based on an affinity model. For crowded scenarios, frequently occurring interactions between objects complicate associations, and affinity-based methods usually fail in these scenarios. Here we introduce quadratic pseudo-Boolean optimization (QPBO) to an online MOT model to analyze frequent interactions. Specifically, we formulate two useful interaction types as pairwise potentials in QPBO, a design that benefits our model by exploiting informative interactions and allowing our online tracker to handle complex scenes. The auxiliary interactions result in a non-submodular QPBO, so we accelerate our online tracker by solving the model with a graph cut combined with a simple heuristic method. This combination achieves a reasonable local optimum and, importantly, implements the tracker efficiently. Extensive experiments on publicly available datasets from both static and moving cameras demonstrate the superiority of our method.

IS Journal 2016 Journal Article

Pattern Recognition, Part 1 [Guest editors' introduction]

• Cheng-Lin Liu
• Brian Lovell
• Dacheng Tao
• Massimo Tistarelli

This special issue reports the advances in pattern recognition theory and applications, particularly, the basic issues of pattern classification and image analysis and their applications. The selected articles address image feature representation, contextual pattern analysis, compact low-rank sparse representation for abnormal event detection in video, customer churn prediction using ensemble learning, online text-independent writer identification using deep convolutional neural network, image depth ordering reasoning, and brain MR image tumor segmentation, respectively.

IS Journal 2016 Journal Article

Pattern Recognition, Part 2

• Cheng-Lin Liu
• Brian Lovell
• Dacheng Tao
• Massimo Tistarelli

This second part of the special issue on pattern recognition reports the advances in pattern recognition for visual data. The selected articles address visual categorization by cross-domain dictionary learning, facial expression recognition, face sketch-photo matching, heartbeat rate measurement from facial video, driver gaze estimation, nighttime vehicle detection, and overlaid arrow detection in biomedical images, respectively.

AAAI Conference 2016 Conference Paper

Random Mixed Field Model for Mixed-Attribute Data Restoration

• Qiang Li
• Wei Bian
• Richard Xu
• Jane You
• Dacheng Tao

Noisy and incomplete data restoration is a critical preprocessing step in developing effective learning algorithms, which targets to reduce the effect of noise and missing values in data. By utilizing attribute correlations and/or instance similarities, various techniques have been developed for data denoising and imputation tasks. However, current existing data restoration methods are either specifically designed for a particular task, or incapable of dealing with mixed-attribute data. In this paper, we develop a new probabilistic model to provide a general and principled method for restoring mixed-attribute data. The main contributions of this study are twofold: a) a unified generative model, utilizing a generic random mixed field (RMF) prior, is designed to exploit mixedattribute correlations; and b) a structured mean-field variational approach is proposed to solve the challenging inference problem of simultaneous denoising and imputation. We evaluate our method by classification experiments on both synthetic data and real benchmark datasets. Experiments demonstrate, our approach can effectively improve the classification accuracy of noisy and incomplete data by comparing with other data restoration methods.

AAAI Conference 2016 Conference Paper

Scalable Completion of Nonnegative Matrix with Separable Structure

• Xiyu Yu
• Wei Bian
• Dacheng Tao

Matrix completion is to recover missing/unobserved values of a data matrix from very limited observations. Due to widely potential applications, it has received growing interests in fields from machine learning, data mining, to collaborative filtering and computer vision. To ensure the successful recovery of missing values, most existing matrix completion algorithms utilise the low-rank assumption, i. e. , the fully observed data matrix has a low rank, or equivalently the columns of the matrix can be linearly represented by a few numbers of basis vectors. Although such low-rank assumption applies generally in practice, real-world data can process much richer structural information. In this paper, we present a new model for matrix completion, motivated by the separability assumption of nonnegative matrices from the recent literature of matrix factorisations: there exists a set of columns of the matrix such that the resting columns can be represented by their convex combinations. Given the separability property, which holds reasonably for many applications, our model provides a more accurate matrix completion than the low-rank based algorithms. Further, we derives a scalable algorithm to solve our matrix completion model, which utilises a randomised method to select the basis columns under the separability assumption and a coordinate gradient based method to automatically deal with the structural constraints in optimisation. Compared to the state-of-the-art algorithms, the proposed matrix completion model achieves competitive results on both synthetic and real datasets.

AAAI Conference 2016 Conference Paper

Shakeout: A New Regularized Deep Neural Network Training Scheme

• Guoliang Kang
• Jun Li
• Dacheng Tao

Recent years have witnessed the success of deep neural networks in dealing with a plenty of practical problems. The invention of effective training techniques largely contributes to this success. The so-called "Dropout" training scheme is one of the most powerful tool to reduce over-fitting. From the statistic point of view, Dropout works by implicitly imposing an L2 regularizer on the weights. In this paper, we present a new training scheme: Shakeout. Instead of randomly discarding units as Dropout does at the training stage, our method randomly chooses to enhance or inverse the contributions of each unit to the next layer. We show that our scheme leads to a combination of L1 regularization and L2 regularization imposed on the weights, which has been proved effective by the Elastic Net models in practice. We have empirically evaluated the Shakeout scheme and demonstrated that sparse network weights are obtained via Shakeout training. Our classification experiments on real-life image datasets MNIST and CIFAR- 10 show that Shakeout deals with over-fitting effectively.

AAAI Conference 2016 Conference Paper

Submodular Asymmetric Feature Selection in Cascade Object Detection

• Baosheng Yu
• Meng Fang
• Dacheng Tao
• Jie Yin

A cascade classifier has turned out to be effective in slidingwindow based real-time object detection. In a cascade classifier, node learning is the key process, which includes feature selection and classifier design. Previous algorithms fail to effectively tackle the asymmetry and intersection problems existing in cascade classification, thereby limiting the performance of object detection. In this paper, we improve current feature selection algorithm by addressing both asymmetry and intersection problems. We formulate asymmetric feature selection as a submodular function maximization problem. We then propose a new algorithm SAFS with formal performance guarantee to solve this problem. We use face detection as a case study and perform experiments on two real-world face detection datasets. The experimental results demonstrate that our algorithm SAFS outperforms the state-of-art feature selection algorithms in cascade object detection, such as FFS and LACBoost.

AAAI Conference 2016 Conference Paper

Teaching-to-Learn and Learning-to-Teach for Multi-label Propagation

• Chen Gong
• Dacheng Tao
• Jie Yang
• Wei Liu

Multi-label propagation aims to transmit the multi-label information from labeled examples to unlabeled examples based on a weighted graph. Existing methods ignore the specific propagation difficulty of different unlabeled examples and conduct the propagation in an imperfect sequence, leading to the error-prone classification of some difficult examples with uncertain labels. To address this problem, this paper associates each possible label with a “teacher”, and proposes a “Multi-Label Teaching-to-Learn and Learning-to- Teach” (ML-TLLT) algorithm, so that the entire propagation process is guided by the teachers and manipulated from simple examples to more difficult ones. In the teaching-to-learn step, the teachers select the simplest examples for the current propagation by investigating both the definitiveness of each possible label of the unlabeled examples, and the dependencies between labels revealed by the labeled examples. In the learning-to-teach step, the teachers reversely learn from the learner’s feedback to properly select the simplest examples for the next propagation. Thorough empirical studies show that due to the optimized propagation sequence designed by the teachers, ML-TLLT yields generally better performance than seven state-of-the-art methods on the typical multi-label benchmark datasets.

TIST Journal 2016 Journal Article

Video Face Editing Using Temporal-Spatial-Smooth Warping

• Xiaoyan Li
• Tongliang Liu
• Jiankang Deng
• Dacheng Tao

Editing faces in videos is a popular yet challenging task in computer vision and graphics that encompasses various applications, including facial attractiveness enhancement, makeup transfer, face replacement, and expression manipulation. Directly applying the existing warping methods to video face editing has the major problem of temporal incoherence in the synthesized videos, which cannot be addressed by simply employing face tracking techniques or manual interventions, as it is difficult to eliminate the subtly temporal incoherence of the facial feature point localizations in a video sequence. In this article, we propose a temporal-spatial-smooth warping (TSSW) method to achieve a high temporal coherence for video face editing. TSSW is based on two observations: (1) the control lattices are critical for generating warping surfaces and achieving the temporal coherence between consecutive video frames, and (2) the temporal coherence and spatial smoothness of the control lattices can be simultaneously and effectively preserved. Based upon these observations, we impose the temporal coherence constraint on the control lattices on two consecutive frames, as well as the spatial smoothness constraint on the control lattice on the current frame. TSSW calculates the control lattice (in either the horizontal or vertical direction) by updating the control lattice (in the corresponding direction) on its preceding frame, i.e., minimizing a novel energy function that unifies a data-driven term, a smoothness term, and feature point constraints. The contributions of this article are twofold: (1) we develop TSSW, which is robust to the subtly temporal incoherence of the facial feature point localizations and is effective to preserve the temporal coherence and spatial smoothness of the control lattices for editing faces in videos, and (2) we present a new unified video face editing framework that is capable for improving the performances of facial attractiveness enhancement, makeup transfer, face replacement, and expression manipulation.

ICML Conference 2015 Conference Paper

Discovering Temporal Causal Relations from Subsampled Data

• Mingming Gong
• Kun Zhang 0001
• Bernhard Schölkopf
• Dacheng Tao
• Philipp Geiger

Granger causal analysis has been an important tool for causal analysis for time series in various fields, including neuroscience and economics, and recently it has been extended to include instantaneous effects between the time series to explain the contemporaneous dependence in the residuals. In this paper, we assume that the time series at the true causal frequency follow the vector autoregressive model. We show that when the data resolution becomes lower due to subsampling, neither the original Granger causal analysis nor the extended one is able to discover the underlying causal relations. We then aim to answer the following question: can we estimate the temporal causal relations at the right causal frequency from the subsampled data? Traditionally this suffers from the identifiability problems: under the Gaussianity assumption of the data, the solutions are generally not unique. We prove that, however, if the noise terms are non-Gaussian, the underlying model for the high frequency data is identifiable from subsampled data under mild conditions. We then propose an Expectation-Maximization (EM) approach and a variational inference approach to recover temporal causal relations from such subsampled data. Experimental results on both simulated and real data are reported to illustrate the performance of the proposed approaches.

TIST Journal 2015 Journal Article

Exploring Spatial Correlation for Visual Object Retrieval

• Miaojing Shi
• Xinghai Sun
• Dacheng Tao
• Chao Xu
• George Baciu
• Hong Liu

Bag-of-visual-words (BOVW)-based image representation has received intense attention in recent years and has improved content-based image retrieval (CBIR) significantly. BOVW does not consider the spatial correlation between visual words in natural images and thus biases the generated visual words toward noise when the corresponding visual features are not stable. This article outlines the construction of a visual word co-occurrence matrix by exploring visual word co-occurrence extracted from small affine-invariant regions in a large collection of natural images. Based on this co-occurrence matrix, we first present a novel high-order predictor to accelerate the generation of spatially correlated visual words and a penalty tree (PTree) to continue generating the words after the prediction. Subsequently, we propose two methods of co-occurrence weighting similarity measure for image ranking: Co-Cosine and Co-TFIDF. These two new schemes down-weight the contributions of the words that are less discriminative because of frequent co-occurrences with other words. We conduct experiments on Oxford and Paris Building datasets, in which the ImageNet dataset is used to implement a large-scale evaluation. Cross-dataset evaluations between the Oxford and Paris datasets and Oxford and Holidays datasets are also provided. Thorough experimental results suggest that our method outperforms the state of the art without adding much additional cost to the BOVW model.

AAAI Conference 2015 Conference Paper

Large-Margin Multi-Label Causal Feature Learning

• Chang Xu
• Dacheng Tao
• Chao Xu

In multi-label learning, an example is represented by a descriptive feature associated with several labels. Simply considering labels as independent or correlated is crude; it would be beneficial to define and exploit the causality between multiple labels. For example, an image label ‘lake’ implies the label ‘water’, but not vice versa. Since the original features are a disorderly mixture of the properties originating from different labels, it is intuitive to factorize these raw features to clearly represent each individual label and its causality relationship. Following the large-margin principle, we propose an effective approach to discover the causal features of multiple labels, thus revealing the causality between labels from the perspective of feature. We show theoretically that the proposed approach is a tight approximation of the empirical multi-label classification error, and the causality revealed strengthens the consistency of the algorithm. Extensive experimentations using synthetic and real-world data demonstrate that the proposed algorithm effectively discovers label causality, generates causal features, and improves multi-label learning.

AAAI Conference 2015 Conference Paper

Low-Rank Multi-View Learning in Matrix Completion for Multi-Label Image Classification

• Meng Liu
• Yong Luo
• Dacheng Tao
• Chao Xu
• Yonggang Wen

Multi-label image classification is of significant interest due to its major role in real-world web image analysis applications such as large-scale image retrieval and browsing. Recently, matrix completion (MC) has been developed to deal with multi-label classification tasks. MC has distinct advantages, such as robustness to missing entries in the feature and label spaces and a natural ability to handle multi-label problems. However, current MC-based multi-label image classi- fication methods only consider data represented by a singleview feature, therefore, do not precisely characterize images that contain several semantic concepts. An intuitive way to utilize multiple features taken from different views is to concatenate the different features into a long vector; however, this concatenation is prone to over-fitting and leads to high time complexity in MC-based image classification. Therefore, we present a novel multi-view learning model for MCbased image classification, called low-rank multi-view matrix completion (lrMMC), which first seeks a low-dimensional common representation of all views by utilizing the proposed low-rank multi-view learning (lrMVL) algorithm. In lrMVL, the common subspace is constrained to be low rank so that it is suitable for MC. In addition, combination weights are learned to explore complementarity between different views. An efficient solver based on fixed-point continuation (FPC) is developed for optimization, and the learned low-rank representation is then incorporated into MC-based image classification. Extensive experimentation on the challenging PAS- CAL VOC’ 07 dataset demonstrates the superiority of lr- MMC compared to other multi-label image classification approaches.

IJCAI Conference 2015 Conference Paper

Multi-Task Model and Feature Joint Learning

• Ya Li
• Xinmei Tian
• Tongliang Liu
• Dacheng Tao

Given several tasks, multi-task learning (MTL) learns multiple tasks jointly by exploring the interdependence between them. The basic assumption in MTL is that those tasks are indeed related. Existing MTL methods model the task relatedness/interdependence in two different ways, either common parameter-sharing or common featuresharing across tasks. In this paper, we propose a novel multi-task learning method to jointly learn shared parameters and shared feature representation. Our objective is to learn a set of common features with which the tasks are related as closely as possible, therefore common parameters shared across tasks can be optimally learned. We present a detailed deviation of our multi-task learning method and propose an alternating algorithm to solve the non-convex optimization problem. We further present a theoretical bound which directly demonstrates that the proposed multi-task learning method can successfully model the relatedness via joint common parameter- and common featurelearning. Extensive experiments are conducted on several real world multi-task learning datasets. All results demonstrate the effectiveness of our multitask model and feature joint learning method.

IJCAI Conference 2015 Conference Paper

Multi-View Matrix Decomposition: A New Scheme for Exploring Discriminative Information

• Cheng Deng
• Zongting Lv
• Wei Liu
• Junzhou Huang
• Dacheng Tao
• Xinbo Gao

Recent studies have demonstrated the advantages of fusing information from multiple views for various machine learning applications. However, most existing approaches assumed the shared component common to all views and ignored the private components of individual views, which thereby restricts the learning performance. In this paper, we propose a new multi-view, low-rank, and sparse matrix decomposition scheme to seamlessly integrate diverse yet complementary information stemming from multiple views. Unlike previous approaches, our approach decomposes an input data matrix concatenated from multiple views as the sum of lowrank, sparse, and noisy parts. Then a unified optimization framework is established, where the lowrankness and group-structured sparsity constraints are imposed to simultaneously capture the shared and private components in both instance and view levels. A proven optimization algorithm is developed to solve the optimization, yielding the learned augmented representation which is used as features for classification tasks. Extensive experiments conducted on six benchmark image datasets show that our approach enjoys superior performance over the state-of-the-art approaches.

IJCAI Conference 2015 Conference Paper

Multi-view Self-Paced Learning for Clustering

• Chang Xu
• Dacheng Tao
• Chao Xu

Exploiting the information from multiple views can improve clustering accuracy. However, most existing multi-view clustering algorithms are nonconvex and are thus prone to becoming stuck into bad local minima, especially when there are outliers and missing data. To overcome this problem, we present a new multi-view self-paced learning (MSPL) algorithm for clustering, that learns the multi-view model by not only progressing from ‘easy’ to ‘complex’ examples, but also from ‘easy’ to ‘complex’ views. Instead of binarily separating the examples or views into ‘easy’ and ‘complex’, we design a novel probabilistic smoothed weighting scheme. Employing multiple views for clustering and defining complexity across both examples and views are shown theoretically to be beneficial to optimal clustering. Experimental results on toy and real-world data demonstrate the efficacy of the proposed algorithm.

TIST Journal 2015 Journal Article

Nonnegative Multiresolution Representation-Based Texture Image Classification

• Yongsheng Dong
• Dacheng Tao
• Xuelong Li

Effective representation of image texture is important for an image-classification task. Statistical modelling in wavelet domains has been widely used to image texture representation. However, due to the intraclass complexity and interclass diversity of textures, it is hard to use a predefined probability distribution function to fit adaptively all wavelet subband coefficients of different textures. In this article, we propose a novel modelling approach, Heterogeneous and Incrementally Generated Histogram (HIGH), to indirectly model the wavelet coefficients by use of four local features in wavelet subbands. By concatenating all the HIGHs in all wavelet subbands of a texture, we can construct a nonnegative multiresolution vector (NMV) to represent a texture image. Considering the NMV’s high dimensionality and nonnegativity, we further propose a Hessian regularized discriminative nonnegative matrix factorization to compute a low-dimensional basis of the linear subspace of NMVs. Finally, we present a texture classification approach by projecting NMVs on the low-dimensional basis. Experimental results show that our proposed texture classification method outperforms seven representative approaches.

TIST Journal 2015 Journal Article

Where2Stand

• Yinting Wang
• Mingli Song
• Dacheng Tao
• Yong Rui
• Jiajun Bu
• Ah Chung Tsoi
• Shaojie Zhuo
• Ping Tan

People often take photographs at tourist sites and these pictures usually have two main elements: a person in the foreground and scenery in the background. This type of “souvenir photo” is one of the most common photos clicked by tourists. Although algorithms that aid a user-photographer in taking a well-composed picture of a scene exist [Ni et al. 2013], few studies have addressed the issue of properly positioning human subjects in photographs. In photography, the common guidelines of composing portrait images exist. However, these rules usually do not consider the background scene. Therefore, in this article, we investigate human-scenery positional relationships and construct a photographic assistance system to optimize the position of human subjects in a given background scene, thereby assisting the user in capturing high-quality souvenir photos. We collect thousands of well-composed portrait photographs to learn human-scenery aesthetic composition rules. In addition, we define a set of negative rules to exclude undesirable compositions. Recommendation results are achieved by combining the first learned positive rule with our proposed negative rules. We implement the proposed system on an Android platform in a smartphone. The system demonstrates its efficacy by producing well-composed souvenir photos.

AAAI Conference 2014 Conference Paper

Active Learning for Crowdsourcing Using Knowledge Transfer

• Meng Fang
• Jie Yin
• Dacheng Tao

This paper studies the active learning problem in crowdsourcing settings, where multiple imperfect annotators with varying levels of expertise are available for labeling the data in a given task. Annotations collected from these labelers may be noisy and unreliable, and the quality of labeled data needs to be maintained for data mining tasks. Previous solutions have attempted to estimate individual users’ reliability based on existing knowledge in each task, but for this to be effective each task requires a large quantity of labeled data to provide accurate estimates. In practice, annotation budgets for a given task are limited, so each instance can be presented to only a few users, each of whom can only label a few examples. To overcome data scarcity we propose a new probabilistic model that transfers knowledge from abundant unlabeled data in auxiliary domains to help estimate labelers’ expertise. Based on this model we present a novel active learning algorithm that: a) simultaneously selects the most informative example and b) queries its label from the labeler with the best expertise. Experiments on both text and image datasets demonstrate that our proposed method outperforms other stateof-the-art active learning methods.

ICML Conference 2014 Conference Paper

Large-margin Weakly Supervised Dimensionality Reduction

• Chang Xu 0002
• Dacheng Tao
• Chao Xu 0006
• Yong Rui

This paper studies dimensionality reduction in a weakly supervised setting, in which the preference relationship between examples is indicated by weak cues. A novel framework is proposed that integrates two aspects of the large margin principle (angle and distance), which simultaneously encourage angle consistency between preference pairs and maximize the distance between examples in preference pairs. Two specific algorithms are developed: an alternating direction method to learn a linear transformation matrix and a gradient boosting technique to optimize a non-linear transformation directly in the function space. Theoretical analysis demonstrates that the proposed large margin optimization criteria can strengthen and improve the robustness and generalization performance of preference learning algorithms on the obtained low-dimensional subspace. Experimental results on real-world datasets demonstrate the significance of studying dimensionality reduction in the weakly supervised setting and the effectiveness of the proposed framework.

AAAI Conference 2014 Conference Paper

ReLISH: Reliable Label Inference via Smoothness Hypothesis

• Chen Gong
• Dacheng Tao
• Keren Fu
• Jie Yang

The smoothness hypothesis is critical for graph-based semi-supervised learning. This paper defines local smoothness, based on which a new algorithm, Reliable Label Inference via Smoothness Hypothesis (ReLISH), is proposed. ReLISH has produced smoother labels than some existing methods for both labeled and unlabeled examples. Theoretical analyses demonstrate good stability and generalizability of ReLISH. Using real-world datasets, our empirical analyses reveal that ReLISH is promising for both transductive and inductive tasks, when compared with representative algorithms, including Harmonic Functions, Local and Global Consistency, Constraint Metric Learning, Linear Neighborhood Propagation, and Manifold Regularization.

AAAI Conference 2014 Conference Paper

Signed Laplacian Embedding for Supervised Dimension Reduction

• Chen Gong
• Dacheng Tao
• Jie Yang
• Keren Fu

Manifold learning is a powerful tool for solving nonlinear dimension reduction problems. By assuming that the high-dimensional data usually lie on a low-dimensional manifold, many algorithms have been proposed. However, most algorithms simply adopt the traditional graph Laplacian to encode the data locality, so the discriminative ability is limited and the embedding results are not always suitable for the subsequent classification. Instead, this paper deploys the signed graph Laplacian and proposes Signed Laplacian Embedding (SLE) for supervised dimension reduction. By exploring the label information, SLE comprehensively transfers the discrimination carried by the original data to the embedded low-dimensional space. Without perturbing the discrimination structure, SLE also retains the locality. Theoretically, we prove the immersion property by computing the rank of projection, and relate SLE to existing algorithms in the frame of patch alignment. Thorough empirical studies on synthetic and real datasets demonstrate the effectiveness of SLE.

IJCAI Conference 2013 Conference Paper

A Bayesian Factorised Covariance Model for Image Analysis

• Jun Li
• Dacheng Tao

This paper presents a specialised Bayesian model for analysing the covariance of data that are observed in the form of matrices, which is particularly suitable for images. Compared to existing generalpurpose covariance learning techniques, we exploit the fact that the variables are organised as an array with two sets of ordered indexes, which induces innate relationship between the variables. Specifically, we adopt a factorised structure for the covariance matrix. The covariance of two variables is represented by the product of the covariance of the two corresponding rows and that of the two columns. The factors, i. e. the row-wise and column-wise covariance matrices are estimated by Bayesian inference with sparse priors. Empirical study has been conducted on image analysis. The model first learns correlations between the rows and columns in an image plane. Then the correlations between individual pixels can be inferred by their locations. This scheme utilises the structural information of an image, and benefits the analysis when the data are damaged or insufficient.

YNIMG Journal 2013 Journal Article

Limited relationships between two-year changes in sulcal morphology and other common neuroimaging indices in the elderly

• Tao Liu
• Perminder S. Sachdev
• Darren M. Lipnicki
• Jiyang Jiang
• Guangqiang Geng
• Wanlin Zhu
• Simone Reppermund
• Dacheng Tao

Measuring the geometry or morphology of sulcal folds has recently become an important approach to investigating neuroanatomy. However, relationships between cortical sulci and other brain structures are poorly understood. The present study investigates how age-related changes in sulcal width are associated with age-related changes in traditional indices of brain structure such as cortical thickness, and cortical gray matter (GM), white matter (WM), subcortical, and white matter hyperintensity (WMH) volumes. These indices and sulcal width were measured at baseline and at two-year follow up in 185 community-dwelling individuals (91 men) aged 70–89years. There were significant increases in sulcal width and WMH volume, and significant decreases in all other indices between baseline and follow-up. Sulcal widening was associated with decreases in cortical GM, subcortical and WM volumes. A further association between sulcal width and cortical thickness became non-significant when cortical GM volume was controlled for. Our findings give insights into the mechanisms responsible for cortical sulcal morphology. The relationships between sulcal morphology and other common measures suggest that it could be a more comprehensive measure for clinical classifications than traditional neuroimaging metrics, such as cortical thickness.

IJCAI Conference 2013 Conference Paper

Semi-Supervised Learning with Manifold Fitted Graphs

• Tongtao Zhang
• Rongrong Ji
• Wei Liu
• Dacheng Tao
• Gang Hua

In this paper, we propose a locality-constrained and sparsity-encouraged manifold fitting approach, aiming at capturing the locally sparse manifold structure into neighborhood graph construction by exploiting a principled optimization model. The proposed model formulates neighborhood graph construction as a sparse coding problem with the locality constraint, therefore achieving simultaneous neighbor selection and edge weight optimization. The core idea underlying our model is to perform a sparse manifold fitting task for each data point so that close-by points lying on the same local manifold are automatically chosen to connect and meanwhile the connection weights are acquired by simple geometric reconstruction. We term the novel neighborhood graph generated by our proposed optimization model M-Fitted Graph since such a graph stems from sparse manifold fitting. To evaluate the robustness and effectiveness of M-fitted graphs, we leverage graph-based semisupervised learning as the testbed. Extensive experiments carried out on six benchmark datasets validate that the proposed M-fitted graph is superior to state-of-the-art neighborhood graphs in terms of classification accuracy using popular graph-based semi-supervised learning methods.

IJCAI Conference 2013 Conference Paper

Shifted Subspaces Tracking on Sparse Outlier for Motion Segmentation

• Tianyi Zhou
• Dacheng Tao

In low-rank & sparse matrix decomposition, the entries of the sparse part are often assumed to be i. i. d. sampled from a random distribution. But the structure of sparse part, as the central interest of many problems, has been rarely studied. One motivating problem is tracking multiple sparse object flows (motions) in video. We introduce “shifted subspaces tracking (SST)” to segment the motions and recover their trajectories by exploring the low-rank property of background and the shifted subspace property of each motion. SST is composed of two steps, background modeling and flow tracking. In step 1, we propose “semi-soft GoDec” to separate all the motions from the low-rank background L as a sparse outlier S. Its soft-thresholding in updating S significantly speeds up GoDec and facilitates the parameter tuning. In step 2, we update X as S obtained in step 1 and develop “SST algorithm” further decomposing X as X = Pk i=1 L(i)◦τ(i)+ S+G, wherein L(i) is a low-rank matrix storing the ith flow after transformation τ(i). SST algorithm solves k sub-problems in sequel by alternating minimization, each of which recovers one L(i) and its τ(i) by randomized method. Sparsity of L(i) and between-frame affinity are leveraged to save computations. We justify the effectiveness of SST on surveillance video sequences.

AAAI Conference 2013 Conference Paper

Supervised Nonnegative Tensor Factorization with Maximum-Margin Constraint

• Fei Wu
• Xu Tan
• Yi Yang
• Dacheng Tao
• Siliang Tang
• Yueting Zhuang

Non-negative tensor factorization (NTF) has attracted great attention in the machine learning community. In this paper, we extend traditional non-negative tensor factorization into a supervised discriminative decomposition, referred as Supervised Non-negative Tensor Factorization with Maximum-Margin Constraint (SNTFM2 ). SNTFM2 formulates the optimal discriminative factorization of non-negative tensorial data as a coupled least-squares optimization problem via a maximum-margin method. As a result, SNTFM2 not only faithfully approximates the tensorial data by additive combinations of the basis, but also obtains a strong generalization power to discriminative analysis (in particular for classification in this paper). The experimental results show the superiority of our proposed model over state-of-the-art techniques on both toy and real world data sets.

AAAI Conference 2013 Conference Paper

Vector-Valued Multi-View Semi-Supervsed Learning for Multi-Label Image Classification

• Yong Luo
• Dacheng Tao
• Chang Xu
• Dongchen Li
• Chao Xu

Images are usually associated with multiple labels and comprised of multiple views, due to each image containing several objects (e. g. a pedestrian, bicycle and tree) and multiple visual features (e. g. color, texture and shape). Currently available tools tend to use either labels or features for classification, but both are necessary to describe the image properly. There have been recent successes in using vector-valued functions, which construct matrix-valued kernels, to explore the multi-label structure in the output space. This has motivated us to develop multi-view vector-valued manifold regularization (MV3 MR) in order to integrate multiple features. MV3 MR exploits the complementary properties of different features, and discovers the intrinsic local geometry of the compact support shared by different features, under the theme of manifold regularization. We validate the effectiveness of the proposed MV3 MR methodology for image classification by conducting extensive experiments on two challenge datasets, PASCAL VOC’ 07 and MIR Flickr.

IROS Conference 2012 Conference Paper

Clustering-based discriminative locality alignment for face gender recognition

• Duo Chen 0001
• Jun Cheng 0002
• Dacheng Tao

To facilitate human-robot interactions, human gender information is very important. Motivated by the success of manifold learning for visual recognition, we present a novel clustering-based discriminative locality alignment (CDLA) algorithm to discover the low-dimensional intrinsic submanifold from the embedding high-dimensional ambient space for improving the face gender recognition performance. In particular, CDLA exploits the global geometry through k-means clustering, extracts the discriminative information through margin maximization and explores the local geometry through intra cluster sample concentration. These three properties uniquely characterize CDLA for face gender recognition. The experimental results obtained from the FERET data sets suggest the superiority of the proposed method in terms of recognition speed and accuracy by comparing with several representative methods.

ICML Conference 2011 Conference Paper

GoDec: Randomized Lowrank & Sparse Matrix Decomposition in Noisy Case

• Tianyi Zhou 0001
• Dacheng Tao

IJCAI Conference 2011 Conference Paper

Learning a Distance Metric by Empirical Loss Minimization

• Wei Bian
• Dacheng Tao

In this paper, we study the problem of learning ametric and propose a loss function based metriclearning framework, in which the metric is estimatedby minimizing an empirical risk over a trainingset. With mild conditions on the instance distributionand the used loss function, we prove that theempirical risk converges to its expected counterpartat rate O(1/\sqrt{n}), wherein n is the cardinality of the training set. In addition, with the assumption thatthe best metric that minimizes the expected risk isbounded, we prove that the learned metric is consistent. Two example algorithms are presented by usingthe proposed loss function based metric learningframework, each of which uses a log loss functionand a smoothed hinge loss function, respectively. Experimental results on data sets from the UCI machine learning repository suggest the effectivenessof the proposed algorithms.

UAI Conference 2011 Conference Paper

Risk Bounds for Infinitely Divisible Distribution

• Chao Zhang 0017
• Dacheng Tao

In this paper, we study the risk bounds for samples independently drawn from an infinitely divisible (ID) distribution. In particular, based on a martingale method, we develop two deviation inequalities for a sequence of random variables of an ID distribution with zero Gaussian component. By applying the deviation inequalities, we obtain the risk bounds based on the covering number for the ID distribution. Finally, we analyze the asymptotic convergence of the risk bound derived from one of the two deviation inequalities and show that the convergence rate of the bound is faster than the result for the generic i.i.d. empirical process (Mendelson, 2003).

AAAI Conference 2010 Conference Paper

Constrained Metric Learning Via Distance Gap Maximization

• Wei Liu
• Xinmei Tian
• Dacheng Tao
• Jianzhuang Liu

Vectored data frequently occur in a variety of fields, which are easy to handle since they can be mathematically abstracted as points residing in a Euclidean space. An appropriate distance metric in the data space is quite demanding for a great number of applications. In this paper, we pose robust and tractable metric learning under pairwise constraints that are expressed as similarity judgements between data pairs. The major features of our approach include: 1) it maximizes the gap between the average squared distance among dissimilar pairs and the average squared distance among similar pairs; 2) it is capable of propagating similar constraints to all data pairs; and 3) it is easy to implement in contrast to the existing approaches using expensive optimization such as semidefinite programming. Our constrained metric learning approach has widespread applicability without being limited to particular backgrounds. Quantitative experiments are performed for classification and retrieval tasks, uncovering the effectiveness of the proposed approach.

NeurIPS Conference 2009 Conference Paper

Manifold Regularization for SIR with Rate Root-n Convergence

• Wei Bian
• Dacheng Tao

In this paper, we study the manifold regularization for the Sliced Inverse Regression (SIR). The manifold regularization improves the standard SIR in two aspects: 1) it encodes the local geometry for SIR and 2) it enables SIR to deal with transductive and semi-supervised learning problems. We prove that the proposed graph Laplacian based regularization is convergent at rate root-n. The projection directions of the regularized SIR are optimized by using a conjugate gradient method on the Grassmann manifold. Experimental results support our theory.