Min Zhang

EAAI Journal 2026 Journal Article

A dual-response colorimetric sensor array integrated with deep learning for mobile intelligent freshness assessment of aquatic products

• Qi Yu
• Min Zhang
• Dayuan Wang
• Chenlin Wu
• Chung Lim Law

The growing demand for rapid, on-site monitoring of aquatic product freshness has highlighted the limitations of conventional methods, which often rely on sophisticated instruments and time-consuming procedures. This study presents an engineering-integrated system for rapid freshness assessment of aquatic products (salmon and shrimp), combining a dual-response colorimetric sensor array with systematically evaluated off-the-shelf deep learning models. In terms of engineering, the sensor array adopts a triple-channel design incorporating pH-responsive anthocyanin, alizarin red S, and indole-specific p-dimethylaminobenzaldehyde, enabling simultaneous detection of total volatile basic nitrogen and indole metabolites during spoilage. On the artificial intelligence front, six representative deep learning architectures were benchmarked on this task, with GhostNet and Xception achieving classification accuracy exceeding 98. 00%. By integrating dual-response signals, the system captures complementary spoilage pathways, improving average accuracy from 95. 90% (single-channel) to 97. 13%, demonstrating the algorithmic of dual-response data fusion. Furthermore, for engineering application, the optimized MobileNet_v1 model was successfully deployed in a mobile application, enabling real-time freshness detection with an accuracy of 97. 20% and an inference time of only 12 ms. This work establishes a reliable framework for on-site food quality monitoring, offering a cost-effective and promising alternative to conventional methods while enhancing supply chain transparency.

FM Conference 2026 Conference Paper

A Formal Framework for Predicting Distributed System Performance Under Faults

• Ziwei Zhou
• Si Liu
• Zhou Zhou
• Peixin Wang
• Min Zhang

Abstract Today’s distributed systems operate in complex environments that inevitably involve faults and even adversarial behaviors. Predicting their performance under such environments directly from formal designs remains a long-standing challenge. We present the first formal framework that systematically enables performance prediction of distributed systems across diverse faulty scenarios. Our framework features a fault injector together with a wide range of faults, reusable as a library, and model compositions that integrate the system and the fault injector into a unified model suitable for statistical analysis of performance properties such as throughput and latency. We formalize the framework in Maude and implement it as an automated tool, PerF. Applied to representative distributed systems, PerF accurately predicts system performance under varying fault settings, with estimations from formal designs consistent with evaluations on real deployments.

AAAI Conference 2026 Conference Paper

Auto-PRE: An Automatic and Cost-Efficient Peer-Review Framework for Language Generation Evaluation

• Junjie Chen
• Weihang Su
• Zhumin Chu
• Haitao Li
• Yujia Zhou
• Dingbo Yuan
• Xudong Wang
• Jun Zhou

The rapid development of large language models (LLMs) has highlighted the need for efficient and reliable methods to evaluate their performance. Traditional evaluation methods often face challenges like high costs, limited task formats, dependence on human references, and systematic biases. To address these limitations, we propose Auto-PRE, an automatic LLM evaluation framework inspired by the peer review process. Unlike previous approaches that rely on human annotations, Auto-PRE automatically selects evaluator LLMs based on three core traits: consistency, pertinence, and self-confidence, which correspond to the instruction, content, and response stages, respectively, and collectively cover the entire evaluation process. Experiments on three representative tasks, including summarization, non-factoid QA, and dialogue generation, demonstrate that Auto-PRE achieves state-of-the-art performance while significantly reducing evaluation costs. Furthermore, the structured and scalable design of our automatic qualification exam framework provides valuable insights into automating the evaluation of LLMs-as-judges, paving the way for more advanced LLM-based evaluation frameworks.

AAAI Conference 2026 Conference Paper

Collaborative LLM Numerical Reasoning with Local Data Protection

• Min Zhang
• Yuzhe Lu
• Yun Zhou
• Panpan Xu
• Lin Lee Cheong
• Chang-Tien Lu
• Haozhu Wang

Numerical reasoning over documents, which demands both contextual understanding and logical inference, is challenging for low-capacity local models deployed on computation-constrained devices. Although such complex reasoning queries could be routed to powerful remote models like GPT-4, exposing local data raises significant data leakage concerns. Existing mitigation methods generate problem descriptions or examples for remote assistance. However, the inherent complexity of numerical reasoning hinders the local model from generating logically equivalent queries and accurately inferring answers with remote guidance. In this paper, we present a model collaboration framework with two key innovations: (1) a context-aware synthesis strategy that shifts the query topics while preserving reasoning patterns; and (2) a tool-based answer reconstruction approach that reuses the remote-generated plug-and-play solution with code snippets. Experimental results demonstrate that our method achieves better reasoning accuracy than solely using local models while providing stronger data protection than fully relying on remote models. Furthermore, our method improves accuracy by 16.2% - 43.6% while reducing data leakage by 2.3% - 44.6% compared to existing data protection approaches.

AAAI Conference 2026 Conference Paper

Improving Value-based Process Verifier via Low-Cost Variance Reduction

• Zetian Sun
• Dongfang Li
• Baotian Hu
• Min Zhang

Large language models (LLMs) have achieved remarkable success in a wide range of tasks. However, their reasoning capabilities, particularly in complex domains like mathematics, remain a significant challenge. Value-based process verifiers, which estimate the probability of a partial reasoning chain leading to a correct solution, are a promising approach for improving reasoning. Nevertheless, their effectiveness is often hindered by estimation error in their training annotations, a consequence of the limited number of Monte Carlo (MC) samples feasible due to the high cost of LLM inference. In this paper, we identify that the estimation error primarily arises from high variance rather than bias, and the MC estimator is a Minimum Variance Unbiased Estimator (MVUE). To address the problem, we propose the Compound Monte Carlo Sampling (ComMCS) method, which constructs an unbiased estimator by linearly combining the MC estimators from the current and subsequent steps. Theoretically, we show that our method leads to a predictable reduction in variance, while maintaining an unbiased estimation without additional LLM inference cost. We also perform empirical experiments on the MATH-500 and GSM8K benchmarks to demonstrate the effectiveness of our method. Notably, ComMCS outperforms regression-based optimization method by 2.8 points, the non-variance-reduced baseline by 2.2 points on MATH-500 on Best-of-32 sampling experiment.

AAAI Conference 2026 Conference Paper

Knowledge Completes the Vision: A Multimodal Entity-aware Retrieval-Augmented Generation Framework for News Image Captioning

• Xiaoxing You
• Qiang Huang
• Lingyu Li
• Chi Zhang
• Xiaopeng Liu
• Min Zhang
• Jun Yu

News image captioning aims to produce journalistically informative descriptions by combining visual content with contextual cues from associated articles. Despite recent advances, existing methods struggle with three key challenges: (1) incomplete information coverage, (2) weak cross-modal alignment, and (3) suboptimal visual-entity grounding. To address these issues, we introduce MERGE, the first Multimodal Entity-aware Retrieval-augmented GEneration framework for news image captioning. MERGE constructs an entity-centric multimodal knowledge base (EMKB) that integrates textual, visual, and structured knowledge, enabling enriched background retrieval. It improves cross-modal alignment through a multistage hypothesis-caption strategy and enhances visual-entity matching via dynamic retrieval guided by image content. Extensive experiments on GoodNews and NYTimes800k show that MERGE significantly outperforms state-of-the-art baselines, with CIDEr gains of +6.84 and +1.16 in caption quality, and F1-score improvements of +4.14 and +2.64 in named entity recognition. Notably, MERGE also generalizes well to the unseen Visual News dataset, achieving +20.17 in CIDEr and +6.22 in F1-score, demonstrating strong robustness and domain adaptability.

AAAI Conference 2026 Conference Paper

PLaST: Towards Paralinguistic-aware Speech Translation

• Yi Li
• Rui Zhao
• Ruiquan Zhang
• Jinsong Su
• Daimeng Wei
• Min Zhang
• Yidong Chen

Speech translation (ST) aims to translate speech from a source language into text in the target language. Naturally, speech signals contain paralinguistic cues beyond linguistic content, which could influence or even alter the interpretation of a lexically identical sentence, thereby yielding distinct translations. However, existing ST models lack direct and sufficient modeling of paralinguistic information, which limits their ability to perceive paralinguistic cues and understand speech comprehensively, leading to degraded translation performance. In response, we propose Paralinguistic-aware Speech Translation (PLaST), a novel dual-branch framework which directly leverages paralinguistic cues beyond the linguistic content. Specifically, PLaST employs a speech encoder and a style extractor to independently generate linguistic and paralinguistic representations, respectively. To obtain a purified linguistic representation aligned with the text representation, a hierarchical Optimal Transport (OT) is applied on the layer-wise outputs from an LLM decoder. Then, the paralinguistic information is retrieved and refined with an Attention-based Retrieval (AR) module, with the linguistic representation serving as queries to enable joint guidance for semantic understanding and translation generation. PLaST outperforms the strong baseline with an average of 5.0 directional and 4.5 global contrastive likelihood scores on the paralinguistic-sensitive benchmark ContraProST, demonstrating its superior capability in paralinguistic perception. Further experiments on the standard speech translation benchmark CoVoST-2 show that PLaST generalizes well to typical ST scenarios.

AAAI Conference 2026 Conference Paper

Spherical Geometry Diffusion: Generating High-quality 3D Face Geometry via Sphere-anchored Representations

• Junyi Zhang
• Yiming Wang
• Yunhong Lu
• Qichao Wang
• Wenzhe Qian
• Xiaoyin Xu
• David Gu
• Min Zhang

A fundamental challenge in text-to-3D face generation is achieving high-quality geometry. The core difficulty lies in the arbitrary and intricate distribution of vertices in 3D space, making it challenging for existing models to establish clean connectivity and resulting in suboptimal geometry. To address this, our core insight is to simplify the underlying geometric structure by constraining the distribution onto a simple and regular manifold, a topological sphere. Building on this, we first propose the Spherical Geometry Representation, a novel face representation that anchors geometric signals to uniform spherical coordinates. This guarantees a regular point distribution, from which the mesh connectivity can be robustly reconstructed. Critically, this canonical sphere can be seamlessly unwrapped into a 2D map, creating a perfect synergy with powerful 2D generative models. We then introduce Spherical Geometry Diffusion, a conditional diffusion framework built upon this 2D map. It enables diverse and controllable generation by jointly modeling geometry and texture, where the geometry explicitly conditions the texture synthesis process. Our method's effectiveness is demonstrated through its success in a wide range of tasks: text-to-3D generation, face reconstruction, and text-based 3D editing. Extensive experiments show that our approach substantially outperforms existing methods in geometric quality, textual fidelity, and inference efficiency.

AAAI Conference 2026 Conference Paper

Tighter Truncated Rectangular Prism Approximation for RNN Robustness Verification

• Xingqi Lin
• Liangyu Chen
• Min Wu
• Min Zhang
• Zhenbing Zeng

Robustness verification is a promising technique for rigorously proving Recurrent Neural Networks (RNNs) robustly. A key challenge is to over-approximate the nonlinear activation functions with linear constraints, which can transform the verification problem into an efficiently solvable linear programming problem. Existing methods over-approximate the nonlinear parts with linear bounding planes individually, which may cause significant over-estimation and lead to lower verification accuracy. In this paper, in order to tightly enclose the three-dimensional nonlinear surface generated by the Hadamard product, we propose a novel truncated rectangular prism formed by two linear relaxation planes and a refinement-driven method to minimize both its volume and surface area for tighter over-approximation. Based on this approximation, we implement a prototype DeepPrism for RNN robustness verification. The experimental results demonstrate that DeepPrism has significant improvement compared with the state-of-the-art approaches in various tasks of image classification, speech recognition and sentiment analysis.

AAMAS Conference 2026 Conference Paper

Towards A Unified Policy Abstraction Theory and Representation Learning Approach in Markov Decision Processes

• Min Zhang
• Hongyao Tang
• Jianye Hao
• Yan Zheng

In intelligent decision-making systems, how policy is represented and optimized is a fundamental problem. The root challenge stems from the large scale and the high complexity of policy space. Towards a desirable surrogate policy space, recent policy representations in a low-dimensional latent space has revealed its potential in improving both evaluation and optimization of policy. The key question to answer in this line of research is by what criterion the policy space should be abstracted for favorable compression and generalization. However, both the theory of policy abstraction and the method of policy representation learning are under-studied. In this work, we first make efforts to fill the vacancy. First, we propose a unified policy abstraction theory, containing three types of policy abstraction and explaining their partial ordering relationship. Then, we generalize policy abstractions to three policy metrics that quantify the distance between policies. Further, we propose a policy representation learning approach and policy optimization algorithm based on deep metric learning. Our study highlights the importance of policy abstraction theory and representation method, demonstratingtheireffectivenessincompressingpolicyspace, characterizing policy differences, and conveying policy generalization.

AAAI Conference 2026 Conference Paper

Towards Closed-Loop Embodied Empathy Evolution: Probing LLM-Centric Lifelong Empathic Motion Generation in Unseen Scenarios

• Jiawen Wang
• Jingjing Wang
• Tianyang Chen
• Min Zhang
• Guodong Zhou

In the literature, existing human-centric emotional motion generation methods primarily focus on boosting performance within a single scale-fixed dataset, largely neglecting the flexible and scale-increasing motion scenarios (e.g., sports, dance), whereas effectively learning these newly emerging scenarios can significantly enhance the model’s real-world generalization ability. Inspired by this, this paper proposes a new LLM-Centric Lifelong Empathic Motion Generation (L2-EMG) task, which aims to equip LLMs with the capability to continually acquire emotional motion generation knowledge across different unseen scenarios, potentially contributing to building a closed-loop and self-evolving embodied agent equipped with both empathy and intelligence. Further, this paper poses two key challenges in the L2-EMG task, i.e., the emotion decoupling challenge and the scenario adapting challenge. To this end, this paper proposes an Emotion-Transferable and Scenario-Adapted Mixture of Experts (ES-MoE) approach which designs a causal-guided emotion decoupling block and a scenario-adapted expert constructing block to address the two challenges, respectively. Especially, this paper constructs multiple L2-EMG datasets to validate the effectiveness of the ES-MoE approach. Extensive evaluations show that ES-MoE outperforms advanced baselines.

IJCAI Conference 2025 Conference Paper

A Survey on the Feedback Mechanism of LLM-based AI Agents

• Zhipeng Liu
• Xuefeng Bai
• Kehai Chen
• Xinyang Chen
• Xiucheng Li
• Yang Xiang
• Jin Liu
• Hong-Dong Li

Large language models (LLMs) are increasingly being adopted to develop general-purpose AI agents. However, it remains challenging for these LLM-based AI agents to efficiently learn from feedback and iteratively optimize their strategies. To address this challenge, tremendous efforts have been dedicated to designing diverse feedback mechanisms for LLM-based AI agents. To provide a comprehensive overview of this rapidly evolving field, this paper presents a systematic review of these studies, offering a holistic perspective on the feedback mechanisms in LLM-based AI agents. We begin by discussing the construction of LLM-based AI agents, introducing a generalized framework that encapsulates much of the existing work. Next, we delve into the exploration of feedback mechanisms, categorizing them into four distinct types: internal feedback, external feedback, multi-agent feedback, and human feedback. Additionally, we provide an overview of evaluation protocols and benchmarks specifically tailored for LLM-based AI agents. Finally, we highlight the significant challenges and identify potential directions for future studies. The relevant papers are summarized and will be consistently updated at https: //github. com/kevinson7515/Agents-Feedback-Mechanisms.

NeurIPS Conference 2025 Conference Paper

AGC-Drive: A Large-Scale Dataset for Real-World Aerial-Ground Collaboration in Driving Scenarios

• Yunhao Hou
• Bochao Zou
• Min Zhang
• 燃 陈
• Shangdong Yang
• Yanmei Zhang
• Junbao Zhuo
• Siheng Chen

By sharing information across multiple agents, collaborative perception helps autonomous vehicles mitigate occlusions and improve overall perception accuracy. While most previous work focus on vehicle-to-vehicle and vehicle-to-infrastructure collaboration, with limited attention to aerial perspectives provided by UAVs, which uniquely offer dynamic, top-down views to alleviate occlusions and monitor large-scale interactive environments. A major reason for this is the lack of high-quality datasets for aerial-ground collaborative scenarios. To bridge this gap, we present AGC-Drive, the first large-scale real-world dataset for Aerial-Ground Cooperative 3D perception. The data collection platform consists of two vehicles, each equipped with five cameras and one LiDAR sensor, and one UAV carrying a forward-facing camera and a LiDAR sensor, enabling comprehensive multi-view and multi-agent perception. Consisting of approximately 80K LiDAR frames and 360K images, the dataset covers 14 diverse real-world driving scenarios, including urban roundabouts, highway tunnels, and on/off ramps. Notably, 17\% of the data comprises dynamic interaction events, including vehicle cut-ins, cut-outs, and frequent lane changes. AGC-Drive contains 350 scenes, each with approximately 100 frames and fully annotated 3D bounding boxes covering 13 object categories. We provide benchmarks for two 3D perception tasks: vehicle-to-vehicle collaborative perception and vehicle-to-UAV collaborative perception. Additionally, we release an open-source toolkit, including spatiotemporal alignment verification tools, multi-agent visualization systems, and collaborative annotation utilities. The dataset and code are available at https: //github. com/PercepX/AGC-Drive.

NeurIPS Conference 2025 Conference Paper

AgentRecBench: Benchmarking LLM Agent-based Personalized Recommender Systems

• Yu Shang
• Peijie Liu
• Yuwei Yan
• Zijing Wu
• Leheng Sheng
• Yuanqing Yu
• Chumeng Jiang
• An Zhang

The emergence of agentic recommender systems powered by Large Language Models (LLMs) represents a paradigm shift in personalized recommendations, leveraging LLMs’ advanced reasoning and role-playing capabilities to enable autonomous, adaptive decision-making. Unlike traditional recommendation approaches, agentic recommender systems can dynamically gather and interpret user-item interactions from complex environments, generating robust recommendation strategies that generalize across diverse scenarios. However, the field currently lacks standardized evaluation protocols to systematically assess these methods. To address this critical gap, we propose: (1) an interactive textual recommendation simulator incorporating rich user and item metadata and three typical evaluation scenarios (classic, evolving-interest, and cold-start recommendation tasks); (2) a unified modular framework for developing agentic recommender systems; and (3) the first comprehensive benchmark comparing over 10 classical and agentic recommendation methods. Our findings demonstrate the superiority of agentic systems and establish actionable design guidelines for their core components. The benchmark environment has been rigorously validated through an open challenge and remains publicly available with a maintained leaderboard at https: //tsinghua-fib-lab. github. io/AgentSocietyChallenge/pages/overview. html. The benchmark is available at: https: //huggingface. co/datasets/SGJQovo/AgentRecBench.

AAAI Conference 2025 Conference Paper

BERT-Based Code Learning for Exception Localization and Type Prediction

• Chongyu Zhang
• Qiping Tao
• Liangyu Chen
• Min Zhang

Exception handling is crucial but challenging in program development. It needs to identify and handle all potential exceptions within programs to ensure system security and stabilization. Traditional exception handling relies on the expertise and experience of programmers, which often leads to oversights. Therefore, identifying exceptional code and recommending handling solutions are hot research topics with significant practical value. This paper presents a model called CodeHunter for exception localization and type prediction. The model first utilizes BERT-based model to represent code features and then uses Bi-LSTM for sequence labeling to pinpoint exceptional code. Additionally, this model also considers contextual features of the exception code and learns weights for the code within the try block and its context through the self-attention mechanism. Subsequently, it performs exception localization and predicts exception types. We conduct experiments on three different datasets. The results demonstrate that in the task of exception localization, our model can achieve a maximum accuracy of 98.6%, exceeding SOTA baselines by 11.2%. In the task of exception type prediction, our model can surpass the accuracy of SOTA baselines by a maximum of 18.7%, achieving 92.0% Top-1 accuracy. The rationality of techniques used in our model is also proved by the ablation testing. The model is implemented as an IDE plugin for programming convenience.

AAAI Conference 2025 Conference Paper

CMT: A Memory Compression Method for Continual Knowledge Learning of Large Language Models

• Dongfang Li
• Zetian Sun
• Xinshuo Hu
• Baotian Hu
• Min Zhang

Large Language Models (LLMs) need to adapt to the continuous changes in data, tasks, and user preferences. Due to their massive size and the high costs associated with training, LLMs are not suitable for frequent retraining. However, updates are necessary to keep them in sync with rapidly evolving human knowledge. To address these challenges, this paper proposes the Compression Memory Training (CMT) method, an efficient and effective online adaptation framework for LLMs that features robust knowledge retention capabilities. Inspired by human memory mechanisms, CMT compresses and extracts information from new documents to be stored in a memory bank. When answering to queries related to these new documents, the model aggregates these document memories from the memory bank to better answer user questions. The parameters of the LLM itself do not change during training and inference, reducing the risk of catastrophic forgetting. To enhance the encoding, retrieval, and aggregation of memory, we further propose three new general and flexible techniques, including memory-aware objective, self-matching and top-k aggregation. Extensive experiments conducted on three continual learning datasets (i.e., StreamingQA, SQuAD and ArchivalQA) demonstrate that the proposed method improves model adaptability and robustness across multiple base LLMs (e.g., +4.07 EM & +4.19 F1 in StreamingQA with Llama-2-7b).

AAAI Conference 2025 Conference Paper

Cobra: Extending Mamba to Multi-Modal Large Language Model for Efficient Inference

• Han Zhao
• Min Zhang
• Wei Zhao
• Pengxiang Ding
• Siteng Huang
• Donglin Wang

In recent years, applying multi-modal large language models (MLLMs) in various fields has achieved remarkable success. However, as the foundation model for many downstream tasks, MLLMs comprise the well-known Transformer network, which has a less efficient quadratic computation complexity. In this study, we introduce Cobra, a multi-modal large-scale language model built upon a state-space model, which has demonstrated significant potential in efficiently handling long sequences with fast inference and linear scalability concerning sequence length. Specifically, Cobra involves replacing Transformer-based backbone models (e.g., LLaMA or Phi) with pre-trained Mamba language models. We then empirically explore effective strategies for aligning visual and textual modalities and integrating various pre-trained Mamba model variants with visual encoders. Experiments across various multi-modal benchmarks demonstrate that: (i) Cobra performs 3× ∼ 4× faster than the most computationally efficient state-of-the-art methods, e.g., LLaVA-Phi and MobileVLM v2. Additionally, its performance is significantly enhanced thanks to the implementation of linear sequential modeling. (ii) Cobra fine-tunes a small parameter (∼48% of model parameters), leading to a significant improvement in overall performance compared to LLaVA.

NeurIPS Conference 2025 Conference Paper

Efficient Speech Language Modeling via Energy Distance in Continuous Latent Space

• Zhengrui Ma
• Yang Feng
• Chenze Shao
• Fandong Meng
• Jie Zhou
• Min Zhang

We introduce \emph{SLED}, an alternative approach to speech language modeling by encoding speech waveforms into sequences of continuous latent representations and modeling them autoregressively using an energy distance objective. The energy distance offers an analytical measure of the distributional gap by contrasting simulated and target samples, enabling efficient training to capture the underlying continuous autoregressive distribution. By bypassing reliance on residual vector quantization, SLED avoids discretization errors and eliminates the need for the complicated hierarchical architectures common in existing speech language models. It simplifies the overall modeling pipeline while preserving the richness of speech information and maintaining inference efficiency. Empirical results demonstrate that SLED achieves strong performance in both zero-shot and streaming speech synthesis, showing its potential for broader applications in general-purpose speech language models. Demos and code are available at \url{https: //github. com/ictnlp/SLED-TTS}.

NeurIPS Conference 2025 Conference Paper

Exploring the Translation Mechanism of Large Language Models

• Hongbin Zhang
• Kehai Chen
• Xuefeng Bai
• Xiucheng Li
• Yang Xiang
• Min Zhang

While large language models (LLMs) demonstrate remarkable success in multilingual translation, their internal core translation mechanisms, even at the fundamental word level, remain insufficiently understood. To address this critical gap, this work introduces a systematic framework for interpreting the mechanism behind LLM translation from the perspective of computational components. This paper first proposes subspace-intervened path patching for precise, fine-grained causal analysis, enabling the detection of components crucial to translation tasks and subsequently characterizing their behavioral patterns in human-interpretable terms. Comprehensive experiments reveal that translation is predominantly driven by a sparse subset of components: specialized attention heads serve critical roles in extracting source language, translation indicators, and positional features, which are then integrated and processed by specific multi-layer perceptrons (MLPs) into intermediary English-centric latent representations before ultimately yielding the final translation. The significance of these findings is underscored by the empirical demonstration that targeted fine-tuning a minimal parameter subset (<5%) enhances translation performance while preserving general capabilities. This result further indicates that these crucial components generalize effectively to sentence-level translation and are instrumental in elucidating more intricate translation tasks.

UAI Conference 2025 Conference Paper

Fast Calculation of Feature Contributions in Boosting Trees

• Zhongli Jiang
• Min Zhang
• Dabao Zhang

Recently, several fast algorithms have been proposed to decompose predicted value into Shapley values, enabling individualized feature contribution analysis in tree models. While such local decomposition offers valuable insights, it underscores the need for a global evaluation of feature contributions. Although coefficients of determination ($R^2$) allow for comparative assessment of individual features, individualizing $R^2$ is challenged by the underlying quadratic losses. To address this, we propose Q-SHAP, an efficient algorithm that reduces the computational complexity of calculating Shapley values for quadratic losses to polynomial time. Our extensive simulations show that Q-SHAP not only improves computational efficiency but also enhances the accuracy of feature-specific $R^2$ estimates.

NeurIPS Conference 2025 Conference Paper

FastLongSpeech: Enhancing Large Speech-Language Models for Efficient Long-Speech Processing

• Shoutao Guo
• Shaolei Zhang
• Qingkai Fang
• Zhengrui Ma
• Min Zhang
• Yang Feng

The rapid advancement of Large Language Models (LLMs) has spurred significant progress in Large Speech-Language Models (LSLMs), enhancing their capabilities in both speech understanding and generation. While existing LSLMs often concentrate on augmenting speech generation or tackling a diverse array of short-speech tasks, the efficient processing of long-form speech remains a critical yet underexplored challenge. This gap is primarily attributed to the scarcity of long-speech training datasets and the high computational costs associated with long sequences. To address these limitations, we introduce FastLongSpeech, a novel framework designed to extend LSLM capabilities for efficient long-speech processing without necessitating dedicated long-speech training data. FastLongSpeech incorporates an iterative fusion strategy that can compress excessively long-speech sequences into manageable lengths. To adapt LSLMs for long-speech inputs, it introduces a dynamic compression training approach, which exposes the model to short-speech sequences at varying compression ratios, thereby transferring the capabilities of LSLMs to long-speech tasks. To assess the long-speech capabilities of LSLMs, we develop a long-speech understanding benchmark called LongSpeech-Eval. Experiments show that our method exhibits strong performance in both long-speech and short-speech tasks, while greatly improving inference efficiency.

AIJ Journal 2025 Journal Article

Grammar induction from visual, speech and text

• Yu Zhao
• Hao Fei
• Shengqiong Wu
• Meishan Zhang
• Min Zhang
• Tat-Seng Chua

Grammar Induction (GI) seeks to uncover the underlying grammatical rules and linguistic patterns of a language, positioning it as a pivotal research topic within Artificial Intelligence (AI). Although extensive research in GI has predominantly focused on text or other singular modalities, we reveal that GI could significantly benefit from rich heterogeneous signals, such as text, vision, and acoustics. In the process, features from distinct modalities essentially serve complementary roles to each other. With such intuition, this work introduces a novel unsupervised visual-audio-text grammar induction task (named VAT-GI ), to induce the constituent grammar trees from parallel images, text, and speech inputs. Inspired by the fact that language grammar natively exists beyond the texts, we argue that the text has not to be the predominant modality in grammar induction. Thus we further introduce a textless setting of VAT-GI, wherein the task solely relies on visual and auditory inputs. To approach the task, we propose a visual-audio-text inside-outside recursive autoencoder ( VaTiora ) framework, which leverages rich modal-specific and complementary features for effective grammar parsing. Besides, a more challenging benchmark data is constructed to assess the generalization ability of VAT-GI system. Experiments on two benchmark datasets demonstrate that our proposed VaTiora system is more effective in incorporating the various multimodal signals, and also presents new state-of-the-art performance of VAT-GI. Further in-depth analyses are shown to gain a deep understanding of the VAT-GI task and how our VaTiora system advances. Our code and data: https://github.com/LLLogen/VAT-GI/

EAAI Journal 2025 Journal Article

Integrated spectrogram construction method on multi-channel signals for loose particle localization

• Zhigang Sun
• Guofu Zhai
• Min Zhang
• Guotao Wang
• Qi Liang
• Hao Chen

Loose particle is an important factor affecting the reliable operation of sealed electronic equipment. Loose particle localization research provides important reference for loose particle cleaning work, but there are still some problems. In this paper, for the first, the authors introduced the “image + deep learning model” scheme into the loose particle localization research, considered conducting multi-channel signal or image recognition research from the source link, thus proposed an integrated spectrogram construction method on multi-channel signals, and conducted two parts of research on signal groups and pulse groups. In the first part, the signal group that provides feedback on overall state information is converted into four signal spectrograms, horizontal merging, vertical merging, cross merging and linear superposition are performed, and four signal-integrated spectrogram sets are created, multiple deep learning models are trained, and the optimal transformation method and appropriate deep learning model are determined. In the second part, the two-stage variable energy threshold pulse extraction algorithm and noise-filling fused threshold-judgment pulse matching algorithm are newly proposed to extract multiple pulse groups from the signal group. The pulse group that provides feedback on relevant state information is converted into four pulse spectrograms, and the same four transformation methods are performed to create four pulse-integrated spectrogram sets and train deep learning models, then the optimal transformation method and appropriate deep learning model are determined. Comprehensive analysis shows that, the signal-group-to-spectrogram construction method that using vertical merging is the optimal integrated spectrogram construction method, Resnet-50 is a well-performing deep learning model, achieving the highest average classification accuracy of 98. 31% and the highest localization accuracy of 100%, which has significant advantages over existing loose particle localization research. A large number of experiments have demonstrated the necessity and correctness of conducting this study, as well as the feasibility, practicality, and superiority of the proposed method. This study is an important supplement to existing loose particle detection research, which can effectively improve the reliability of sealed electronic equipment in aerospace systems. In theory, this study can provide important references for multi-channel signal processing and recognition research in the field of acoustic emission detection and fault diagnosis.

AAAI Conference 2025 Conference Paper

Knowledge Editing with Dynamic Knowledge Graphs for Multi-Hop Question Answering

• Yifan Lu
• Yigeng Zhou
• Jing Li
• Yequan Wang
• Xuebo Liu
• Daojing He
• Fangming Liu
• Min Zhang

Multi-hop question answering (MHQA) poses a significant challenge for large language models (LLMs) due to the extensive knowledge demands involved. Knowledge editing, which aims to precisely modify the LLMs to incorporate specific knowledge without negatively impacting other unrelated knowledge, offers a potential solution for addressing MHQA challenges with LLMs. However, current solutions struggle to effectively resolve issues of knowledge conflicts. Most parameter-preserving editing methods are hindered by inaccurate retrieval and overlook secondary editing issues, which can introduce noise into the reasoning process of LLMs. In this paper, we introduce KEDKG, a novel knowledge editing method that leverages a dynamic knowledge graph for MHQA, designed to ensure the reliability of answers. KEDKG involves two primary steps: dynamic knowledge graph construction and knowledge graph augmented generation. Initially, KEDKG autonomously constructs a dynamic knowledge graph to store revised information while resolving potential knowledge conflicts. Subsequently, it employs a fine-grained retrieval strategy coupled with an entity and relation detector to enhance the accuracy of graph retrieval for LLM generation. Experimental results on benchmarks show that KEDKG surpasses previous state-of-the-art models, delivering more accurate and reliable answers in environments with dynamic information.

NeurIPS Conference 2025 Conference Paper

Learning to Watermark: A Selective Watermarking Framework for Large Language Models via Multi-Objective Optimization

• Chenrui Wang
• Junyi Shu
• Billy Chiu
• Yu Li
• Saleh Alharbi
• Min Zhang
• Jing Li

The rapid development of LLMs has raised concerns about their potential misuse, leading to various watermarking schemes that typically offer high detectability. However, existing watermarking techniques often face trade-off between watermark detectability and generated text quality. In this paper, we introduce Learning to Watermark (LTW), a novel selective watermarking framework that leverages multi-objective optimization to effectively balance these competing goals. LTW features a lightweight network that adaptively decides when to apply the watermark by analyzing sentence embeddings, token entropy, and current watermarking ratio. Training of the network involves two specifically constructed loss functions that guide the model toward Pareto-optimal solutions, thereby harmonizing watermark detectability and text quality. By integrating LTW with two baseline watermarking methods, our experimental evaluations demonstrate that LTW significantly enhances text quality without compromising detectability. Our selective watermarking approach offers a new perspective for designing watermarks for LLMs and a way to preserve high text quality for watermarks. The code is publicly available at: https: //github. com/fattyray/learning-to-watermark

AAAI Conference 2025 Conference Paper

Look Before You Leap: Enhance Attention and Vigilance Regarding Harmful Content with GuidelineLLM

• Shaoqing Zhang
• Zhuosheng Zhang
• Kehai Chen
• Rongxiang Weng
• Muyun Yang
• Tiejun Zhao
• Min Zhang

Despite being empowered with alignment mechanisms, large language models (LLMs) are increasingly vulnerable to emerging jailbreak attacks that can compromise their alignment mechanisms. This vulnerability poses significant risks to real-world applications. Existing work faces challenges in both training efficiency and generalization capabilities (i.e., Reinforcement Learning from Human Feedback and Red-Teaming). Developing effective strategies to enable LLMs to resist continuously evolving jailbreak attempts represents a significant challenge. To address this challenge, we propose a novel defensive paradigm called GuidelineLLM, which assists LLMs in recognizing queries that may have harmful content. Before LLMs respond to a query, GuidelineLLM first identifies potential risks associated with the query, summarizes these risks into guideline suggestions, and then feeds these guidelines to the responding LLMs. Importantly, our approach eliminates the necessity for additional safety fine-tuning of the LLMs themselves; only the GuidelineLLM requires fine-tuning. This characteristic enhances the general applicability of GuidelineLLM across various LLMs. Experimental results demonstrate that GuidelineLLM can significantly reduce the attack success rate (ASR) against LLM (an average reduction of 34.17% ASR) while maintaining the usefulness of LLM in handling benign queries.

NeurIPS Conference 2025 Conference Paper

MASTER: Enhancing Large Language Model via Multi-Agent Simulated Teaching

• Liang Yue
• Yihong Tang
• Kehai Chen
• Jie Liu
• Min Zhang

Instruction fine-tuning is crucial in NLP tasks, enhancing pretrained models' instruction-following capabilities and task-specific performance. However, obtaining high-quality fine-tuning data for large models is challenging due to data collection difficulties and high production costs. To address this, we propose MASTER, a novel data augmentation method that enriches original data through interactions among multiple agents with varying cognitive levels. We simulate three pedagogically grounded teaching scenarios, leveraging multi-agent conversations to generate high-quality teacher-student interaction data. Utilizing MASTER, we construct BOOST-QA, a fine-tuning dataset augmented from existing datasets like Orca-Math-200k, ProcQA, and OpenHermes2. 5. Experiments show that models fine-tuned with BOOST-QA perform excellently across multiple benchmarks, demonstrating strong multitask generalization. Notably, MASTER significantly improves models' reasoning abilities in complex tasks, providing valuable insights for future research.

EAAI Journal 2025 Journal Article

Monitoring the deterioration state of braking friction performance of high-speed train under imbalanced data

• Min Zhang
• Wenxin Xu
• Jiliang Mo
• Zaiyu Xiang
• Zhongrong Zhou

To solve the problem of poor friction block classification when the number of samples of different degradation states is imbalanced in the braking process of high-speed trains, a friction block deterioration state monitoring model based on generative adversarial network (GAN) and lightweight network is proposed in this study. First, the vibration signal of the friction block is transformed into a two-dimensional heat map to enhance the distinguishability between different types of signals. Second, the proposed method is used to expand the samples under different friction blocks. The modified MobileNetv3 (MMN3) lightweight network framework is then used to construct the monitoring model of the friction block degradation state, reducing the number of parameters while maintaining high precision recognition performance. Finally, the proposed method is applied to the imbalanced brake pad friction block data obtained from the self-developed high-speed train test bench. The proposed model can maintain high recognition accuracy in different degrees of friction block wear state imbalance proportion datasets. It can effectively monitor the deterioration state of high-speed train friction blocks under imbalanced data.

EAAI Journal 2025 Journal Article

Multi-source information fused loose particle localization and material identification method for sealed electronic equipment

• Zhigang Sun
• Qi Liang
• Guofu Zhai
• Guotao Wang
• Min Zhang
• Jingting Sun

Sealed electronic equipment are1 an important component of aerospace defense systems, and loose particles pose a significant threat to their reliable operation. Loose particle detection is crucial. For sealed electronic equipment with large scale and complex structure, loose particle detection should not only include the judgment of existence, but also obtain location and material information to facilitate the cleaning and control work. In this paper, the authors proposed a multi-source information fused loose particle localization and material identification method. Firstly, the equipment model was designed, the loose particle samples were made, and loose particle signals were collected. Secondly, the two-stage adaptive energy threshold pulse extraction algorithm was newly proposed to extract effective pulses, and the threshold-judgement-search pulse matching algorithm was improved to match the effective pulse groups. Next, spectrograms were transformed from effective pulses to create the localization and material image set. The time-domain, frequency-domain and gray-level co-occurrence matrix features were used to construct the joint feature library. Then, the channel-weighting feature selection method was used to create the localization and material data set. Finally, PReLU-VGG19-Plus was trained on the localization and material image set to obtain the optimal localization and material neural network, while parameter-optimized XGBoost was trained on the localization and material data set to obtain the optimal localization and material classifier. On this basis, combined with the triple majority voting process, the combined localization and material framework were constructed. Extensive test results show that, the location-identification achieved by combined localization framework and the material-identification accuracy achieved by combined material framework are all 100%. The feasibility, stability, and superiority of the method proposed in this paper have been fully verified. It is an important supplement to the existing loose particle detection research, providing important reference for signal detection and classification research in similar fields, and effectively improving the reliability of sealed electronic equipment.

NeurIPS Conference 2025 Conference Paper

SCAN: Self-Denoising Monte Carlo Annotation for Robust Process Reward Learning

• Yuyang Ding
• Xinyu Shi
• Juntao Li
• Xiaobo Liang
• Zhaopeng Tu
• Min Zhang

Process reward models (PRMs) offer fine-grained, step-level evaluations that facilitate deeper reasoning processes in large language models (LLMs), proving effective in complex tasks like mathematical reasoning. However, developing PRMs is challenging due to the high cost and limited scalability of human-annotated data. Synthetic data from Monte Carlo (MC) estimation is a promising alternative but suffers from a high noise ratio, which can cause overfitting and hinder large-scale training. In this work, we conduct a preliminary study on the noise distribution in synthetic data from MC estimation, identifying that annotation models tend to both underestimate and overestimate step correctness due to limitations in their annotation capabilities. Building on these insights, we propose {\bf S}elf-Denoising Monte {\bf C}arlo {\bf An}notation (\textsc{Scan}), an efficient data synthesis and noise-tolerant learning framework. Our key findings indicate that: (1) Even lightweight models (e. g. , 1. 5B parameters) can produce high-quality annotations through self-denoising strategy, enabling PRMs to achieve superior performance with only 6\% the inference cost required by vanilla MC estimation. (2) With our robust learning strategy, PRMs can effectively learn from this weak supervision, achieving a 39. 2 F1 score improvement (from 19. 9 to 59. 1) in ProcessBench. Despite using only a compact synthetic dataset, our models surpass strong baselines, including those trained on large-scale human-annotated datasets such as PRM800K. Furthermore, performance continues to improve as we scale up the synthetic data, highlighting the potential of \textsc{Scan} for scalable, cost-efficient, and robust PRM training.

NeurIPS Conference 2025 Conference Paper

SSRB: Direct Natural Language Querying to Massive Heterogeneous Semi-Structured Data

• Xin Zhang
• Mingxin Li
• Yanzhao Zhang
• Dingkun Long
• Yongqi Li
• Yinghui Li
• Pengjun Xie
• Meishan Zhang

Searching over semi-structured data with natural language (NL) queries has attracted sustained attention, enabling broader audiences to access information easily. As more applications, such as LLM agents and RAG systems, emerge to search and interact with semi-structured data, two major challenges have become evident: (1) the increasing diversity of domains and schema variations, making domain-customized solutions prohibitively costly; (2) the growing complexity of NL queries, which combine both exact field matching conditions and fuzzy semantic requirements, often involving multiple fields and implicit reasoning. These challenges make formal language querying or keyword-based search insufficient. In this work, we explore neural retrievers as a unified non-formal querying solution by directly index semi-structured collections and understand NL queries. We employ LLM-based automatic evaluation and build a large-scale semi-structured retrieval benchmark (SSRB) using LLM generation and filtering, containing 14M semi-structured objects from 99 different schemas across 6 domains, along with 8, 485 test queries that combine both exact and fuzzy matching conditions. Our systematic evaluation of popular retrievers shows that current state-of-the-art models could achieve acceptable performance, yet they still lack precise understanding of matching constraints. While by in-domain training of dense retrievers, the performance can be significantly improved. We believe that our SSRB could serve as a valuable resource for future research in this area, and we hope to inspire further exploration of semi-structured retrieval with complex queries.

NeurIPS Conference 2025 Conference Paper

Thinking in Character: Advancing Role-Playing Agents with Role-Aware Reasoning

• Yihong Tang
• Kehai Chen
• Muyun Yang
• Zheng-Yu Niu
• Jing Li
• Tiejun Zhao
• Min Zhang

The advancement of Large Language Models (LLMs) has spurred significant interest in Role-Playing Agents (RPAs) for applications such as emotional companionship and virtual interaction. However, recent RPAs are often built on explicit dialogue data, lacking deep, human-like internal thought processes, resulting in superficial knowledge and style expression. While Large Reasoning Models (LRMs) can be employed to simulate character thought, their direct application is hindered by attention diversion (i. e. , RPAs forget their role) and style drift (i. e. , overly formal and rigid reasoning rather than character-consistent reasoning). To address these challenges, this paper introduces a novel Role-Aware Reasoning (RAR) method, which consists of two important stages: Role Identity Activation (RIA) and Reasoning Style Optimization (RSO). RIA explicitly guides the model with character profiles during reasoning to counteract attention diversion, and then RSO aligns reasoning style with the character and scene via LRM distillation to mitigate style drift. Extensive experiments demonstrate that the proposed RAR significantly enhances the performance of RPAs by effectively addressing attention diversion and style drift.

NeurIPS Conference 2025 Conference Paper

VimoRAG: Video-based Retrieval-augmented 3D Motion Generation for Motion Language Models

• Haidong Xu
• Guangwei Xu
• Zhedong Zheng
• Xiatian Zhu
• Wei Ji
• Xiangtai Li
• Ruijie Guo
• Meishan Zhang

This paper introduces VimoRAG, a novel video-based retrieval-augmented motion generation framework for motion large language models (LLMs). As motion LLMs face severe out-of-domain/out-of-vocabulary issues due to limited annotated data, VimoRAG leverages large-scale in-the-wild video databases to enhance 3D motion generation by retrieving relevant 2D human motion signals. While video-based motion RAG is nontrivial, we address two key bottlenecks: (1) developing an effective motion-centered video retrieval model that distinguishes human poses and actions, and (2) mitigating the issue of error propagation caused by suboptimal retrieval results. We design the Gemini Motion Video Retriever mechanism and the Motion-centric Dual-alignment DPO Trainer, enabling effective retrieval and generation processes. Experimental results show that VimoRAG significantly boosts the performance of motion LLMs constrained to text-only input.

NeurIPS Conference 2025 Conference Paper

XIFBench: Evaluating Large Language Models on Multilingual Instruction Following

• Zhenyu Li
• Kehai Chen
• Yunfei Long
• Xuefeng Bai
• Yaoyin Zhang
• Xuchen Wei
• Juntao Li
• Min Zhang

Large Language Models (LLMs) have demonstrated remarkable instruction-following capabilities across various applications. However, their performance in multilingual settings lacks systematic investigation, with existing evaluations lacking fine-grained constraint analysis across diverse linguistic contexts. We introduce XIFBench, a comprehensive constraint-based benchmark for evaluating multilingual instruction-following abilities of LLMs, comprising 558 instructions with 0-5 additional constraints across five categories ( Content, Style, Situation, Format, and Numerical ) in six languages spanning different resource levels. To support reliable and consistent cross-lingual evaluation, we implement three methodological innovations: cultural accessibility annotation, constraint-level translation validation, and requirement-based evaluation using English requirements as semantic anchors across languages. Extensive experiments with various LLMs not only quantify performance disparities across resource levels but also provide detailed insights into how language resources, constraint categories, instruction complexity, and cultural specificity influence multilingual instruction-following. Our code and data are available at https: //github. com/zhenyuli801/XIFBench.

NeurIPS Conference 2024 Conference Paper

$\texttt{ConflictBank}$: A Benchmark for Evaluating the Influence of Knowledge Conflicts in LLMs

• Zhaochen Su
• Jun Zhang
• Xiaoye Qu
• Tong Zhu
• Yanshu Li
• Jiashuo Sun
• Juntao Li
• Min Zhang

Large language models (LLMs) have achievedimpressive advancements across numerous disciplines, yet the critical issue of knowledge conflicts, a major source of hallucinations, has rarely been studied. While a few research explored the conflicts between the inherent knowledge of LLMs and the retrieved contextual knowledge, a comprehensive assessment of knowledge conflict in LLMs is still missing. Motivated by this research gap, we firstly propose ConflictBank, the largest benchmark with 7. 45M claim-evidence pairs and 553k QA pairs, addressing conflicts from misinformation, temporal discrepancies, and semantic divergences. Using ConflictBank, we conduct the thorough and controlled experiments for a comprehensive understanding of LLM behavior in knowledge conflicts, focusing on three key aspects: (i) conflicts encountered in retrieved knowledge, (ii) conflicts within the models' encoded knowledge, and (iii) the interplay between these conflict forms. Our investigation delves into four model families and twelve LLM instances and provides insights into conflict types, model sizes, and the impact at different stages. We believe that knowledge conflicts represent a critical bottleneck to achieving trustworthy artificial intelligence and hope our work will offer valuable guidance for future model training and development. Resources are available at https: //github. com/zhaochen0110/conflictbank.

AAAI Conference 2024 Conference Paper

An Empirical Study of CLIP for Text-Based Person Search

• Min Cao
• Yang Bai
• Ziyin Zeng
• Mang Ye
• Min Zhang

Text-based Person Search (TBPS) aims to retrieve the person images using natural language descriptions. Recently, Contrastive Language Image Pretraining (CLIP), a universal large cross-modal vision-language pre-training model, has remarkably performed over various cross-modal downstream tasks due to its powerful cross-modal semantic learning capacity. TPBS, as a fine-grained cross-modal retrieval task, is also facing the rise of research on the CLIP-based TBPS. In order to explore the potential of the visual-language pre-training model for downstream TBPS tasks, this paper makes the first attempt to conduct a comprehensive empirical study of CLIP for TBPS and thus contribute a straightforward, incremental, yet strong TBPS-CLIP baseline to the TBPS community. We revisit critical design considerations under CLIP, including data augmentation and loss function. The model, with the aforementioned designs and practical training tricks, can attain satisfactory performance without any sophisticated modules. Also, we conduct the probing experiments of TBPS-CLIP in model generalization and model compression, demonstrating the effectiveness of TBPS-CLIP from various aspects. This work is expected to provide empirical insights and highlight future CLIP-based TBPS research.

NeurIPS Conference 2024 Conference Paper

AutoSurvey: Large Language Models Can Automatically Write Surveys

• Yidong Wang
• Qi Guo
• Wenjin Yao
• Hongbo Zhang
• Xin Zhang
• Zhen Wu
• Meishan Zhang
• Xinyu Dai

This paper introduces AutoSurvey, a speedy and well-organized methodology for automating the creation of comprehensive literature surveys in rapidly evolving fields like artificial intelligence. Traditional survey paper creation faces challenges due to the vast volume and complexity of information, prompting the need for efficient survey methods. While large language models (LLMs) offer promise in automating this process, challenges such as context window limitations, parametric knowledge constraints, and the lack of evaluation benchmarks remain. AutoSurvey addresses these challenges through a systematic approach that involves initial retrieval and outline generation, subsection drafting by specialized LLMs, integration and refinement, and rigorous evaluation and iteration. Our contributions include a comprehensive solution to the survey problem, a reliable evaluation method, and experimental validation demonstrating AutoSurvey's effectiveness.

EAAI Journal 2024 Journal Article

Broiler health monitoring technology based on sound features and random forest

• Zhigang Sun
• Weige Tao
• Mengmeng Gao
• Min Zhang
• Shoulai Song
• Guotao Wang

The existing broiler health monitoring technology has problems such as low automation, unstable monitoring results, and low practical value, making it difficult to provide timely and reliable broiler health monitoring results. The broiler sound signal can provide feedback on their health. A widely validated and correct experience is to analyze the frequency of coughs in a segment of broiler sound signal to determine the health of the broiler group. Based on this, in this paper, the authors proposed a new broiler health monitoring technology based on sound detection. The broiler health monitoring problem is cleverly transformed into a multi-classification problem, which can be solved by identifying the sound types in broiler sound signals. Specifically, the audio signal collection system was designed to complete signal collection and preliminary signal filtering. Wiener filtering was used for deep signal filtering. The 60-dimensional sound features with good performance from three aspects, time-frequency domain, Mel-Frequency Cepstral Coefficients, and sparse representation were extracted, and a preliminary data set was created. Min-max normalization was used to align the numerical distribution of the data set, and a high-quality data set was created. Multi-classification models based on different classification algorithms and neural networks were trained, and the best-performing Random Forest was obtained, thus parameter optimization was carried out, and the optimal multi-classification model was obtained, achieving a classification accuracy of 91. 14%. The visualization platform was built to process the classification results of the multi-classification model, completing majority voting processing and cough rate calculation, thereby achieving broiler health monitoring. In addition, the definitions of cough rate and prediction accuracy were newly proposed. A large number of experiments have verified the feasibility of the broiler health monitoring technology proposed in this paper, with an average prediction accuracy of 98. 97% achieved.

JAIR Journal 2024 Journal Article

Cross-domain Constituency Parsing by Leveraging Heterogeneous Data

• Peiming Guo
• Meishan Zhang
• Yulong Chen
• Jianling Li
• Min Zhang
• Yue Zhang

Knowledge transfer is investigated in various natural language processing tasks except cross-domain constituency parsing. In this paper, we leverage heterogeneous data to transfer cross-domain and cross-task knowledge to constituency parsing. Concretely, we first select language modeling, named entity recognition, CCG supertagging and dependency parsing as auxiliary tasks and collect the corpora of these tasks covering various domains as cross-domain and cross-task heterogeneous data. Second, we exploit three types of prefixes: shared, task and domain prefix, to merge cross-domain and cross-task data and decompose the general, task and domain representation in the pretrained language model. Third, we convert the data formats of multi-source heterogeneous datasets and loss objectives of the auxiliary tasks into a consistent formalization closer to constituency parsing. Finally, we jointly train the model to transfer task and domain knowledge to cross-domain constituency parsing. We verify the effectiveness of our proposed model on five target domains of MCTB. Experimental results show that our knowledge transfer model outperforms various baseline models, including conventional chart-based and transition-based parsers and the current large-scale language model for zero-shot and few-shot settings.

AAAI Conference 2024 Conference Paper

Expressive Forecasting of 3D Whole-Body Human Motions

• Pengxiang Ding
• Qiongjie Cui
• Haofan Wang
• Min Zhang
• Mengyuan Liu
• Donglin Wang

Human motion forecasting, with the goal of estimating future human behavior over a period of time, is a fundamental task in many real-world applications. However, existing works typically concentrate on foretelling the major joints of the human body without considering the delicate movements of the human hands. In practical applications, hand gesture plays an important role in human communication with the real world, and expresses the primary intention of human beings. In this work, we are the first to formulate whole-body human pose forecasting task, which jointly predicts future both body and gesture activities. Correspondingly, we propose a novel Encoding-Alignment-Interaction (EAI) framework that aims to predict both coarse (body joints) and fine-grained (gestures) activities collaboratively, enabling expressive and cross-facilitated forecasting of 3D whole-body human motions. Specifically, our model involves two key constituents: cross-context alignment (XCA) and cross-context interaction (XCI). Considering the heterogeneous information within the whole-body, XCA aims to align the latent features of various human components, while XCI focuses on effectively capturing the context interaction among the human components. We conduct extensive experiments on a newly-introduced large-scale benchmark and achieve state-of-the-art performance. The code is public for research purposes at https://github.com/Dingpx/EAI.

NeurIPS Conference 2024 Conference Paper

In-Context Learning State Vector with Inner and Momentum Optimization

• Dongfang Li
• Zhenyu Liu
• Xinshuo Hu
• Zetian Sun
• Baotian Hu
• Min Zhang

Large Language Models (LLMs) have exhibited an impressive ability to perform In-Context Learning (ICL) from only a few examples. Recent works have indicated that the functions learned by ICL can be represented through compressed vectors derived from the transformer. However, the working mechanisms and optimization of these vectors are yet to be thoroughly explored. In this paper, we address this gap by presenting a comprehensive analysis of these compressed vectors, drawing parallels to the parameters trained with gradient descent, and introducing the concept of state vector. Inspired by the works on model soup and momentum-based gradient descent, we propose inner and momentum optimization methods that are applied to refine the state vector progressively as test-time adaptation. Moreover, we simulate state vector aggregation in the multiple example setting, where demonstrations comprising numerous examples are usually too lengthy for regular ICL, and further propose a divide-and-conquer aggregation method to address this challenge. We conduct extensive experiments using Llama-2 and GPT-J in both zero-shot setting and few-shot setting. The experimental results show that our optimization method effectively enhances the state vector and achieves the state-of-the-art performance on diverse tasks.

ICLR Conference 2024 Conference Paper

Large Language Model Cascades with Mixture of Thought Representations for Cost-Efficient Reasoning

• Murong Yue
• Jie Zhao
• Min Zhang
• Liang Du
• Ziyu Yao 0002

Large language models (LLMs) such as GPT-4 have exhibited remarkable performance in a variety of tasks, but this strong performance often comes with the high expense of using paid API services. In this paper, we are motivated to study building an LLM "cascade" to save the cost of using LLMs, particularly for performing (e.g., mathematical, causal) reasoning tasks. Our cascade pipeline follows the intuition that simpler questions can be addressed by a weaker but more affordable LLM, whereas only the most challenging questions necessitate the stronger and more expensive LLM. To realize this decision-making, we consider the "answer consistency" of the weaker LLM as a signal of the question difficulty and propose several methods for answering sampling and consistency checking, including one leveraging a mixture of two thought representations (i.e., Chain-of-Thought and Program-of-Thought). Through experiments on six reasoning benchmark datasets, with GPT-3.5-turbo and GPT-4 being the weaker and stronger LLMs, respectively, our cascade pipeline demonstrates comparable performance but reduces about 60% of the cost compared with fully using the stronger LLM.

NeurIPS Conference 2024 Conference Paper

Learnability Matters: Active Learning for Video Captioning

• Yiqian Zhang
• Buyu Liu
• Jun Bao
• Qiang Huang
• Min Zhang
• Jun Yu

This work focuses on the active learning in video captioning. In particular, we propose to address the learnability problem in active learning, which has been brought up by collective outliers in video captioning and neglected in the literature. To start with, we conduct a comprehensive study of collective outliers, exploring their hard-to-learn property and concluding that ground truth inconsistency is one of the main causes. Motivated by this, we design a novel active learning algorithm that takes three complementary aspects, namely learnability, diversity, and uncertainty, into account. Ideally, learnability is reflected by ground truth consistency. Under the active learning scenario where ground truths are not available until human involvement, we measure the consistency on estimated ground truths, where predictions from off-the-shelf models are utilized as approximations to ground truths. These predictions are further used to estimate sample frequency and reliability, evincing the diversity and uncertainty respectively. With the help of our novel caption-wise active learning protocol, our algorithm is capable of leveraging knowledge from humans in a more effective yet intellectual manner. Results on publicly available video captioning datasets with diverse video captioning models demonstrate that our algorithm outperforms SOTA active learning methods by a large margin, e. g. we achieve about 103% of full performance on CIDEr with 25% of human annotations on MSR-VTT.

AAAI Conference 2024 Conference Paper

Learning to Reweight for Generalizable Graph Neural Network

• Zhengyu Chen
• Teng Xiao
• Kun Kuang
• Zheqi Lv
• Min Zhang
• Jinluan Yang
• Chengqiang Lu
• Hongxia Yang

Graph Neural Networks (GNNs) show promising results for graph tasks. However, existing GNNs' generalization ability will degrade when there exist distribution shifts between testing and training graph data. The fundamental reason for the severe degeneration is that most GNNs are designed based on the I.I.D hypothesis. In such a setting, GNNs tend to exploit subtle statistical correlations existing in the training set for predictions, even though it is a spurious correlation. In this paper, we study the problem of the generalization ability of GNNs on Out-Of-Distribution (OOD) settings. To solve this problem, we propose the Learning to Reweight for Generalizable Graph Neural Network (L2R-GNN) to enhance the generalization ability for achieving satisfactory performance on unseen testing graphs that have different distributions with training graphs. We propose a novel nonlinear graph decorrelation method, which can substantially improve the out-of-distribution generalization ability and compares favorably to previous methods in restraining the over-reduced sample size. The variables of graph representation are clustered based on the stability of their correlations, and graph decorrelation method learns weights to remove correlations between the variables of different clusters rather than any two variables. Besides, we introduce an effective stochastic algorithm based on bi-level optimization for the L2R-GNN framework, which enables simultaneously learning the optimal weights and GNN parameters, and avoids the over-fitting issue. Experiments show that L2R-GNN greatly outperforms baselines on various graph prediction benchmarks under distribution shifts.

NeurIPS Conference 2024 Conference Paper

Neural Collapse Inspired Feature Alignment for Out-of-Distribution Generalization

• Zhikang Chen
• Min Zhang
• Sen Cui
• Haoxuan Li
• Gang Niu
• Mingming Gong
• Changshui Zhang
• Kun Zhang

The spurious correlation between the background features of the image and its label arises due to that the samples labeled with the same class in the training set often co-occurs with a specific background, which will cause the encoder to extract non-semantic features for classification, resulting in poor out-of-distribution generalization performance. Although many studies have been proposed to address this challenge, the semantic and spurious features are still difficult to accurately decouple from the original image and fail to achieve high performance with deep learning models. This paper proposes a novel perspective inspired by neural collapse to solve the spurious correlation problem through the alternate execution of environment partitioning and learning semantic masks. Specifically, we propose to assign an environment to each sample by learning a local model for each environment and using maximum likelihood probability. At the same time, we require that the learned semantic mask neurally collapses to the same simplex equiangular tight frame (ETF) in each environment after being applied to the original input. We conduct extensive experiments on four datasets, and the results demonstrate that our method significantly improves out-of-distribution performance.

NeurIPS Conference 2024 Conference Paper

NewTerm: Benchmarking Real-Time New Terms for Large Language Models with Annual Updates

• Hexuan Deng
• Wenxiang Jiao
• Xuebo Liu
• Min Zhang
• Zhaopeng Tu

Despite their remarkable abilities in various tasks, large language models (LLMs) still struggle with real-time information (e. g. , new facts and terms) due to the knowledge cutoff in their development process. However, existing benchmarks focus on outdated content and limited fields, facing difficulties in real-time updating and leaving new terms unexplored. To address this problem, we propose an adaptive benchmark, NewTerm, for real-time evaluation of new terms. We design a highly automated construction method to ensure high-quality benchmark construction with minimal human effort, allowing flexible updates for real-time information. Empirical results on various LLMs demonstrate over 20% performance reduction caused by new terms. Additionally, while updates to the knowledge cutoff of LLMs can cover some of the new terms, they are unable to generalize to more distant new terms. We also analyze which types of terms are more challenging and why LLMs struggle with new terms, paving the way for future research. Finally, we construct NewTerm 2022 and 2023 to evaluate the new terms updated each year and will continue updating annually. The benchmark and codes can be found at https: //anonymous. 4open. science/r/NewTerms.

NeurIPS Conference 2024 Conference Paper

Parameter Competition Balancing for Model Merging

• Guodong DU
• Junlin Lee
• Jing Li
• Runhua Jiang
• Yifei Guo
• Shuyang Yu
• Hanting Liu
• Sim K. Goh

While fine-tuning pretrained models has become common practice, these models often underperform outside their specific domains. Recently developed model merging techniques enable the direct integration of multiple models, each fine-tuned for distinct tasks, into a single model. This strategy promotes multitasking capabilities without requiring retraining on the original datasets. However, existing methods fall short in addressing potential conflicts and complex correlations between tasks, especially in parameter-level adjustments, posing a challenge in effectively balancing parameter competition across various tasks. This paper introduces an innovative technique named PCB-Merging (Parameter Competition Balancing), a lightweight and training-free technique that adjusts the coefficients of each parameter for effective model merging. PCB-Merging employs intra-balancing to gauge parameter significance within individual tasks and inter-balancing to assess parameter similarities across different tasks. Parameters with low importance scores are dropped, and the remaining ones are rescaled to form the final merged model. We assessed our approach in diverse merging scenarios, including cross-task, cross-domain, and cross-training configurations, as well as out-of-domain generalization. The experimental results reveal that our approach achieves substantial performance enhancements across multiple modalities, domains, model sizes, number of tasks, fine-tuning forms, and large language models, outperforming existing model merging methods.

AAAI Conference 2024 Conference Paper

Prompt-Based Distribution Alignment for Unsupervised Domain Adaptation

• Shuanghao Bai
• Min Zhang
• Wanqi Zhou
• Siteng Huang
• Zhirong Luan
• Donglin Wang
• Badong Chen

Recently, despite the unprecedented success of large pre-trained visual-language models (VLMs) on a wide range of downstream tasks, the real-world unsupervised domain adaptation (UDA) problem is still not well explored. Therefore, in this paper, we first experimentally demonstrate that the unsupervised-trained VLMs can significantly reduce the distribution discrepancy between source and target domains, thereby improving the performance of UDA. However, a major challenge for directly deploying such models on downstream UDA tasks is prompt engineering, which requires aligning the domain knowledge of source and target domains, since the performance of UDA is severely influenced by a good domain-invariant representation. We further propose a Prompt-based Distribution Alignment (PDA) method to incorporate the domain knowledge into prompt learning. Specifically, PDA employs a two-branch prompt-tuning paradigm, namely base branch and alignment branch. The base branch focuses on integrating class-related representation into prompts, ensuring discrimination among different classes. To further minimize domain discrepancy, for the alignment branch, we construct feature banks for both the source and target domains and propose image-guided feature tuning (IFT) to make the input attend to feature banks, which effectively integrates self-enhanced and cross-domain features into the model. In this way, these two branches can be mutually promoted to enhance the adaptation of VLMs for UDA. We conduct extensive experiments on three benchmarks to demonstrate that our proposed PDA achieves state-of-the-art performance. The code is available at https://github.com/BaiShuanghao/Prompt-based-Distribution-Alignment.

EAAI Journal 2024 Journal Article

Research on filtering and classification method for white-feather broiler sound signals based on sparse representation

• Zhigang Sun
• Min Zhang
• Jingang Liu
• Qianyu Wu
• Jiaqi Wang
• Guotao Wang

Aiming at the problems in the existing white-feather broiler sound signal classification research, such as not considering the information and requirements in the real engineering application scenarios, not finding new applicable sound features, not providing the basis for the selection of classification learners, and the problems in the existing white-feather broiler sound signal filtering research, such as not carrying out in-depth research, not considering the signal characteristics, not analyzing the filtering effects in detail. Therefore, in this paper, the authors cleverly combined four methods of signal filtering, signal classification, machine learning, and sparse representation to propose a filtering and classification method for white-feather broiler sound signals based on sparse representation. Specifically, in terms of signal filtering, the signal characteristics of the white-feather broiler sound signal was carefully analyzed, a combination of noise reduction processing and signal reconstruction was newly proposed, and three classic and effective approximation algorithms were adopted, then the optimal filtering method of “using OMPA for noise reduction processing, and using 50-atom OMPA for signal reconstruction” was obtained. In terms of signal classification, in addition to the time-frequency domain and Mel-Frequency Cepstral Coefficients (abbr. MFCCs), time-frequency parameter features from sparse representations were newly extracted, and a total of 60-dimensional sound features were calculated on the frame signals, thus the data set was created. Feature engineering was performed on the data set, finally, the high-quality data set was obtained. Several common machine learning classification algorithms were used to train classification learners on the data set, and random forest with better performance was selected for parameter optimization. The parameter-optimized classification learner achieved the highest average prediction accuracy of 94. 09%. Majority voting was newly proposed to process the prediction results to get the required signal classification result. On this basis, a new definition of classification accuracy suitable for white-feather broiler sound signals was newly proposed. Test results in real engineering application scenarios show that, the classification accuracy achieved by the proposed method on unknown white-feather broiler sound signals is about 88. 89%. Compared with the existing research, the optimal filtering method considers the signal characteristics for the first time and achieves good filtering effects. The trained classification learner obtains excellent prediction effects and achieves good classification effects on several real sound types. This study is an important supplement to existing research on animal sound signal processing and classification, and provides important reference for the subsequent white-feather broiler health monitoring research.

AAAI Conference 2024 Conference Paper

Robustness Verification of Deep Reinforcement Learning Based Control Systems Using Reward Martingales

• Dapeng Zhi
• Peixin Wang
• Cheng Chen
• Min Zhang

Deep Reinforcement Learning (DRL) has gained prominence as an effective approach for control systems. However, its practical deployment is impeded by state perturbations that can severely impact system performance. Addressing this critical challenge requires robustness verification about system performance, which involves tackling two quantitative questions: (i) how to establish guaranteed bounds for expected cumulative rewards, and (ii) how to determine tail bounds for cumulative rewards. In this work, we present the first approach for robustness verification of DRL-based control systems by introducing reward martingales, which offer a rigorous mathematical foundation to characterize the impact of state perturbations on system performance in terms of cumulative rewards. Our verified results provide provably quantitative certificates for the two questions. We then show that reward martingales can be implemented and trained via neural networks, against different types of control policies. Experimental results demonstrate that our certified bounds tightly enclose simulation outcomes on various DRL-based control systems, indicating the effectiveness and generality of the proposed approach.

NeurIPS Conference 2024 Conference Paper

SelectIT: Selective Instruction Tuning for LLMs via Uncertainty-Aware Self-Reflection

• Liangxin Liu
• Xuebo Liu
• Derek F. Wong
• Dongfang Li
• Ziyi Wang
• Baotian Hu
• Min Zhang

Instruction tuning (IT) is crucial to tailoring large language models (LLMs) towards human-centric interactions. Recent advancements have shown that the careful selection of a small, high-quality subset of IT data can significantly enhance the performance of LLMs. Despite this, common approaches often rely on additional models or data, which increases costs and limits widespread adoption. In this work, we propose a novel approach, termed $\textit{SelectIT}$, that capitalizes on the foundational capabilities of the LLM itself. Specifically, we exploit the intrinsic uncertainty present in LLMs to more effectively select high-quality IT data, without the need for extra resources. Furthermore, we introduce a curated IT dataset, the $\textit{Selective Alpaca}$, created by applying SelectIT to the Alpaca-GPT4 dataset. Empirical results demonstrate that IT using Selective Alpaca leads to substantial model ability enhancement. The robustness of SelectIT has also been corroborated in various foundation models and domain-specific tasks. Our findings suggest that longer and more computationally intensive IT data may serve as superior sources of IT, offering valuable insights for future research in this area. Data, code, and scripts are freely available at https: //github. com/Blue-Raincoat/SelectIT.

AAAI Conference 2024 Conference Paper

Separate the Wheat from the Chaff: Model Deficiency Unlearning via Parameter-Efficient Module Operation

• Xinshuo Hu
• Dongfang Li
• Baotian Hu
• Zihao Zheng
• Zhenyu Liu
• Min Zhang

Large language models (LLMs) have been widely used in various applications but are known to suffer from issues related to untruthfulness and toxicity. While parameter-efficient modules (PEMs) have demonstrated their effectiveness in equipping models with new skills, leveraging PEMs for deficiency unlearning remains underexplored. In this work, we propose a PEMs operation approach, namely Extraction-before-Subtraction (Ext-Sub), to enhance the truthfulness and detoxification of LLMs through the integration of ``expert'' PEM and ``anti-expert'' PEM. Remarkably, even anti-expert PEM possess valuable capabilities due to their proficiency in generating fabricated content, which necessitates language modeling and logical narrative competence. Rather than merely negating the parameters, our approach involves extracting and eliminating solely the deficiency capability within anti-expert PEM while preserving the general capabilities. To evaluate the effectiveness of our approach in terms of truthfulness and detoxification, we conduct extensive experiments on LLMs, encompassing additional abilities such as language modelling and mathematical reasoning. Our empirical results demonstrate that our approach effectively improves truthfulness and detoxification, while largely preserving the fundamental abilities of LLMs.

EAAI Journal 2024 Journal Article

Signal detection and material identification method for loose particles inside sealed relays based on fusion classification model

• Zhigang Sun
• Guotao Wang
• Guofu Zhai
• Pengfei Li
• Qi Liang
• Min Zhang

In this study, the authors combined the research on loose particle signal and component signal identification with the research on loose particle material identification for the first time, providing comprehensive and reliable loose particle detection results. Based on this, a signal detection and material identification method for loose particles inside sealed relays based on fusion classification model is proposed. Due to the limitations of technical means and confidentiality management, the authors made a real sealed relay sample, and took it as the research object. Through the steps of data acquisition, signal processing, feature engineering, and model training, the dedicated component identification feature library and material identification feature library was constructed, respectively, the component identification model and material identification model based on parameter-optimized SVM with linear kernel and XGBoost was trained, respectively. For the seal relay to be tested, through the steps of data acquisition, signal processing and feature engineering, the data set to be tested was created. The component identification model was used to identify component signals with loose particle signals, and the material identification model was used to identify the materials of loose particles. The majority voting process was used to convert the classification results into identification results, resulting in loose particle detection and material identification results. In addition, the general procedure steps of the proposed method for physical testing were given, and the identification accuracy for device-level loose particle detection was newly proposed. The loose particle testing event containing thirty-seven identification tasks shows that, the achieved identification accuracy was 97. 30%, and 92. 16% of the average classification accuracy was achieved by the component identification model, 80. 41% of the average classification accuracy was achieved by the material identification model. This effectively demonstrates the feasibility and practicality of the proposed method in this paper. It is an important supplement to the loose particle detection research, and provides references for signal detection in similar fields.

NeurIPS Conference 2024 Conference Paper

Synergistic Dual Spatial-aware Generation of Image-to-text and Text-to-image

• Yu Zhao
• Hao Fei
• Xiangtai Li
• Libo Qin
• Jiayi Ji
• Hongyuan Zhu
• Meishan Zhang
• Min Zhang

In the visual spatial understanding (VSU) field, spatial image-to-text (SI2T) and spatial text-to-image (ST2I) are two fundamental tasks that appear in dual form. Existing methods for standalone SI2T or ST2I perform imperfectly in spatial understanding, due to the difficulty of 3D-wise spatial feature modeling. In this work, we consider modeling the SI2T and ST2I together under a dual learning framework. During the dual framework, we then propose to represent the 3D spatial scene features with a novel 3D scene graph (3DSG) representation that can be shared and beneficial to both tasks. Further, inspired by the intuition that the easier 3D$\to$image and 3D$\to$text processes also exist symmetrically in the ST2I and SI2T, respectively, we propose the Spatial Dual Discrete Diffusion (SD$^3$) framework, which utilizes the intermediate features of the 3D$\to$X processes to guide the hard X$\to$3D processes, such that the overall ST2I and SI2T will benefit each other. On the visual spatial understanding dataset VSD, our system outperforms the mainstream T2I and I2T methods significantly. Further in-depth analysis reveals how our dual learning strategy advances.

NeurIPS Conference 2024 Conference Paper

The Ladder in Chaos: Improving Policy Learning by Harnessing the Parameter Evolving Path in A Low-dimensional Space

• Hongyao Tang
• Min Zhang
• Chen Chen
• Jianye Hao

Knowing the learning dynamics of policy is significant to unveiling the mysteries of Reinforcement Learning (RL). It is especially crucial yet challenging to Deep RL, from which the remedies to notorious issues like sample inefficiency and learning instability could be obtained. In this paper, we study how the policy networks of typical DRL agents evolve during the learning process by empirically investigating several kinds of temporal change for each policy parameter. In popular MuJoCo and DeepMind Control Suite (DMC) environments, we find common phenomena for TD3 and RAD agents: (1) the activity of policy network parameters is highly asymmetric and policy networks advance monotonically along a very limited number of major parameter directions; (2) severe detours occur in parameter update and harmonic-like changes are observed for all minor parameter directions. By performing a novel temporal SVD along the policy learning path, the major and minor parameter directions are identified as the columns of the right unitary matrix associated with dominant and insignificant singular values respectively. Driven by the discoveries above, we propose a simple and effective method, called Policy Path Trimming and Boosting (PPTB), as a general plug-in improvement to DRL algorithms. The key idea of PPTB is to trim the policy learning path by canceling the policy updates in minor parameter directions, and boost the learning path by encouraging the advance in major directions. In experiments, we demonstrate that our method improves the learning performance of TD3, RAD, and DoubleDQN regarding scores and efficiency in MuJoCo, DMC, and MinAtar tasks respectively.

EAAI Journal 2023 Journal Article

A systematic empirical study on word embedding based methods in discovering Chinese black keywords

• Chenyang Wang
• YI Shen
• Yuwei Li
• Min Zhang
• Miao Hu
• Jinghua Zheng

With the development of online transactions, the Chinese cyber black market is proliferating and facilitates many cybercrimes. It is difficult to understand the cyber black market due to the confusing jargon (called black keywords in this paper) used by criminals to conceal underground transactions. To discover black keywords automatically, some natural language processing based methods have been proposed by comparing the similarity of word vectors generated by word embedding models. Therefore, the quality of word vectors generated has a significant impact on black keyword discovery and it is necessary to evaluate different word embedding models in discovering black keywords. To this end, we design a Chinese black keyword discovery framework and conduct a systematic empirical study on six existing word embedding models including both static and dynamic types in discovering Chinese black keywords. In specific, we classify Chinese black keywords in four types: domain specific words (DSWs), new meaning words (NMWs), similar pronunciation words (SPWs), and similar glyph words (SGWs). We experimentally find that different word embedding models vary greatly in performance when discovering black keywords, e. g. , dynamic models perform well in discovering DSWs and NMWs, static ones perform poorly in discovering NMWs. We improve the static word embedding model based NMW discovery algorithm by additionally comparing the differences in cross-corpus word nearest-neighbors before and after domain incremental training. For effectively discovering variant words like SPWs and SGWs, we additionally introduce Chinese pronunciation and glyph features. The experimental results demonstrate the effectiveness of the proposed Chinese black keyword discovery framework, with detection accuracies of over 90% for DSWs, 80% for NWMs, 90% for SPWs, and 61% for SGWs.

AAAI Conference 2023 Conference Paper

Adversarial Self-Attention for Language Understanding

• Hongqiu Wu
• Ruixue Ding
• Hai Zhao
• Pengjun Xie
• Fei Huang
• Min Zhang

Deep neural models (e.g. Transformer) naturally learn spurious features, which create a ``shortcut'' between the labels and inputs, thus impairing the generalization and robustness. This paper advances self-attention mechanism to its robust variant for Transformer-based pre-trained language models (e.g. BERT). We propose Adversarial Self-Attention mechanism (ASA), which adversarially biases the attentions to effectively suppress the model reliance on features (e.g. specific keywords) and encourage its exploration of broader semantics. We conduct comprehensive evaluation across a wide range of tasks for both pre-training and fine-tuning stages. For pre-training, ASA unfolds remarkable performance gain compared to naive training for longer steps. For fine-tuning, ASA-empowered models outweigh naive models by a large margin considering both generalization and robustness.

AAAI Conference 2023 Conference Paper

AMOM: Adaptive Masking over Masking for Conditional Masked Language Model

• Yisheng Xiao
• Ruiyang Xu
• Lijun Wu
• Juntao Li
• Tao Qin
• Tie-Yan Liu
• Min Zhang

Transformer-based autoregressive (AR) methods have achieved appealing performance for varied sequence-to-sequence generation tasks, e.g., neural machine translation, summarization, and code generation, but suffer from low inference efficiency. To speed up the inference stage, many non-autoregressive (NAR) strategies have been proposed in the past few years. Among them, the conditional masked language model (CMLM) is one of the most versatile frameworks, as it can support many different sequence generation scenarios and achieve very competitive performance on these tasks. In this paper, we further introduce a simple yet effective adaptive masking over masking strategy to enhance the refinement capability of the decoder and make the encoder optimization easier. Experiments on 3 different tasks (neural machine translation, summarization, and code generation) with 15 datasets in total confirm that our proposed simple method achieves significant performance improvement over the strong CMLM model. Surprisingly, our proposed model yields state-of-the-art performance on neural machine translation (34.62 BLEU on WMT16 EN to RO, 34.82 BLEU on WMT16 RO to EN, and 34.84 BLEU on IWSLT De to En) and even better performance than the AR Transformer on 7 benchmark datasets with at least 2.2x speedup. Our code is available at GitHub.

AIJ Journal 2023 Journal Article

Are the BERT family zero-shot learners? A study on their potential and limitations

• Yue Wang
• Lijun Wu
• Juntao Li
• Xiaobo Liang
• Min Zhang

Starting from the resurgence of deep learning, language models (LMs) have never been so popular. Through simply increasing model scale and data size, large LMs pre-trained with self-supervision objectives demonstrate awe-inspiring results on both task performance and generalization. At the early stage, supervised fine-tuning is indispensable in adapting pre-trained language models (PLMs) to downstream tasks. Later on, the sustained growth of model capacity and data size, as well as newly presented pre-training techniques, make the PLMs perform well under the few-shot setting, especially in the recent paradigm of prompt-based learning. After witnessing the success of PLMs for few-shot tasks, we propose to further study the potential and limitations of PLMs for the zero-shot setting. We utilize 3 models from the most popular BERT family to launch the empirical study on 20 different datasets. We are surprised to find that some simple strategies (without the need of human efforts or unsupervised data) can yield very promising results on a few widely-used datasets, e. g. , 88. 34 % ( ± 0. 60 ) accuracy on the IMDB dataset, and 84. 88 % ( ± 2. 83 ) accuracy on the Amazon dataset, which outperforms manually created prompts without engineering in achieving much better and stable performance with the accuracy of 74. 06 % ( ± 13. 04 ), 75. 54 % ( ± 11. 77 ) for comparison. However, we also observe some limitations of PLMs under the zero-shot setting, particularly for the language understanding tasks (e. g. , GLUE, SuperGLUE). 2

NeurIPS Conference 2023 Conference Paper

Beyond MLE: Convex Learning for Text Generation

• Chenze Shao
• Zhengrui Ma
• Min Zhang
• Yang Feng

Maximum likelihood estimation (MLE) is a statistical method used to estimate the parameters of a probability distribution that best explain the observed data. In the context of text generation, MLE is often used to train generative language models, which can then be used to generate new text. However, we argue that MLE is not always necessary and optimal, especially for closed-ended text generation tasks like machine translation. In these tasks, the goal of model is to generate the most appropriate response, which does not necessarily require it to estimate the entire data distribution with MLE. To this end, we propose a novel class of training objectives based on convex functions, which enables text generation models to focus on highly probable outputs without having to estimate the entire data distribution. We investigate the theoretical properties of the optimal predicted distribution when applying convex functions to the loss, demonstrating that convex functions can sharpen the optimal distribution, thereby enabling the model to better capture outputs with high probabilities. Experiments on various text generation tasks and models show the effectiveness of our approach. It enables autoregressive models to bridge the gap between greedy and beam search, and facilitates the learning of non-autoregressive models with a maximum improvement of 9+ BLEU points. Moreover, our approach also exhibits significant impact on large language models (LLMs), substantially enhancing their generative capability on various tasks. Source code is available at \url{https: //github. com/ictnlp/Convex-Learning}.

NeurIPS Conference 2023 Conference Paper

Boosting Verification of Deep Reinforcement Learning via Piece-Wise Linear Decision Neural Networks

• Jiaxu Tian
• Dapeng Zhi
• Si Liu
• Peixin Wang
• Cheng Chen
• Min Zhang

Formally verifying deep reinforcement learning (DRL) systems suffers from both inaccurate verification results and limited scalability. The major obstacle lies in the large overestimation introduced inherently during training and then transforming the inexplicable decision-making models, i. e. , deep neural networks (DNNs), into easy-to-verify models. In this paper, we propose an inverse transform-then-train approach, which first encodes a DNN into an equivalent set of efficiently and tightly verifiable linear control policies and then optimizes them via reinforcement learning. We accompany our inverse approach with a novel neural network model called piece-wise linear decision neural networks (PLDNNs), which are compatible with most existing DRL training algorithms with comparable performance against conventional DNNs. Our extensive experiments show that, compared to DNN-based DRL systems, PLDNN-based systems can be more efficiently and tightly verified with up to $438$ times speedup and a significant reduction in overestimation. In particular, even a complex $12$-dimensional DRL system is efficiently verified with up to 7 times deeper computation steps.

AAAI Conference 2023 Conference Paper

Deep Attentive Model for Knowledge Tracing

• Xinping Wang
• Liangyu Chen
• Min Zhang

Knowledge Tracing (KT) is a crucial task in the field of online education, since it aims to predict students' performance on exercises based on their learning history. One typical solution for knowledge tracing is to combine the classic models in educational psychology, such as Item Response Theory (IRT) and Cognitive Diagnosis (CD), with Deep Neural Networks (DNN) technologies. In this solution, a student and related exercises are mapped into feature vectors based on the student's performance at the current time step, however, it does not consider the impact of historical behavior sequences, and the relationships between historical sequences and students. In this paper, we develop DAKTN, a novel model which assimilates the historical sequences to tackle this challenge for better knowledge tracing. To be specific, we apply a pooling layer to incorporate the student behavior sequence in the embedding layer. After that, we further design a local activation unit, which can adaptively calculate the representation vectors by taking the relevance of historical sequences into consideration with respect to candidate student and exercises. Through experimental results on three real-world datasets, DAKTN significantly outperforms state-of-the-art baseline models. We also present the reasonableness of DAKTN by ablation testing.

EAAI Journal 2023 Journal Article

Feature optimization method for white feather broiler health monitoring technology

• Weige Tao
• Guotao Wang
• Zhigang Sun
• Shuyan Xiao
• Lingjiao Pan
• Quanyu Wu
• Min Zhang

The existing research on white feather broiler health monitoring technology concentrate on the selection of classification algorithms and optimization of internal parameters, ignoring the white feather broiler sound signal data set (abbr. signal data set) and internal features for training classifiers. In this paper, the authors shifted the research focus to the signal data set, and proposed an applicable feature optimization method in order to improve the practicality and stability of broiler health monitoring technology. First, for outliers in the signal data set, the boxplot was used to detect them, and they were transformed into missing values. For all missing values in the signal data set, six missing value processing methods were used to process them, respectively. The optimal decision tree classifier (abbr. DT classifier) filling was determined. Second, three normalization methods were used to process the signal data set, respectively, and the optimal min–max normalization method was determined. Then, the Pearson correlation coefficient and the p -value were used to perform correlation tests on features and labels in the signal data set, then the linear–nonlinear-fusion-based (abbr. LNLF-based) feature selection method was newly proposed to select those features that have a great influence on the classification effect of the random forest classifier (abbr. RF classifier). Multiple tests had verified its superiority. Finally, ten white feather broiler sound signal validation data sets (abbr. validation data sets) were used to verify the proposed feature optimization method. It shows that the average classification accuracy achieved by the RF classifier on the signal data set before and after feature optimization improved from 72. 85% to 85. 28%, with an improvement of 12. 43%. The average G-mean achieved by the RF classifier improved from 0. 828 to 0. 939, with an improvement of 0. 111. The average classification accuracy and average G-mean achieved by the RF classifier on the validation data set before and after feature optimization improved significantly, with an average improvement of 13. 26% and 0. 093, respectively. This method improves the reliability of the existing classifiers in another way, which is an important supplement to the feature engineering study, and promotes the research progress of the existing white feather broiler health monitoring technology.

YNIMG Journal 2023 Journal Article

Functional connectivity profiles of the default mode and visual networks reflect temporal accumulative effects of sustained naturalistic emotional experience

• Shuyue Xu
• Zhiguo Zhang
• Linling Li
• Yongjie Zhou
• Danyi Lin
• Min Zhang
• Li Zhang
• Gan Huang

Determining and decoding emotional brain processes under ecologically valid conditions remains a key challenge in affective neuroscience. The current functional Magnetic Resonance Imaging (fMRI) based emotion decoding studies are mainly based on brief and isolated episodes of emotion induction, while sustained emotional experience in naturalistic environments that mirror daily life experiences are scarce. Here we used 12 different 10-minute movie clips as ecologically valid emotion-evoking procedures in n = 52 individuals to explore emotion-specific fMRI functional connectivity (FC) profiles on the whole-brain level at high spatial resolution (432 parcellations including cortical and subcortical structures). Employing machine-learning based decoding and cross validation procedures allowed to investigate FC profiles contributing to classification that can accurately distinguish sustained happiness and sadness and that generalize across subjects, movie clips, and parcellations. Both functional brain network-based and subnetwork-based emotion classification results suggested that emotion manifests as distributed representation of multiple networks, rather than a single functional network or subnetwork. Further, the results showed that the Visual Network (VN) and Default Mode Network (DMN) associated functional networks, especially VN-DMN, exhibited a strong contribution to emotion classification. To further estimate the temporal accumulative effect of naturalistic long-term movie-based video-evoking emotions, we divided the 10-min episode into three stages: early stimulation (1∼200 s), middle stimulation (201∼400 s), and late stimulation (401∼600 s) and examined the emotion classification performance at different stimulation stages. We found that the late stimulation contributes most to the classification (accuracy=85.32%, F1-score=85.62%) compared to early and middle stimulation stages, implying that continuous exposure to emotional stimulation can lead to more intense emotions and further enhance emotion-specific distinguishable representations. The present work demonstrated that sustained happiness and sadness under naturalistic conditions are presented in emotion-specific network profiles and these expressions may play different roles in the generation and modulation of emotions. These findings elucidated the importance of network level adaptations for sustained emotional experiences during naturalistic contexts and open new venues for imaging network level contributions under naturalistic conditions.

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.

IJCAI Conference 2023 Conference Paper

Learning Discrete Representations via Constrained Clustering for Effective and Efficient Dense Retrieval (Extended Abstract)

• Jingtao Zhan
• Jiaxin Mao
• Yiqun Liu
• Jiafeng Guo
• Min Zhang
• Shaoping Ma

Dense Retrieval~(DR) has achieved state-of-the-art first-stage ranking effectiveness. However, the efficiency of most existing DR models is limited by the large memory cost of storing dense vectors and the time-consuming nearest neighbor search~(NNS) in vector space. Therefore, we present RepCONC, a novel retrieval model that learns discrete Representations via CONstrained Clustering. RepCONC jointly trains dual-encoders and the Product Quantization~(PQ) method to learn discrete document representations and enables fast approximate NNS with compact indexes. It models quantization as a constrained clustering process, which requires the document embeddings to be uniformly clustered around the quantization centroids. We theoretically demonstrate that the uniform clustering constraint facilitates representation distinguishability. Extensive experiments show that RepCONC substantially outperforms a wide range of existing retrieval models in terms of retrieval effectiveness, memory efficiency, and time efficiency.

EAAI Journal 2023 Journal Article

Method of Locating Loose Particles Inside Aerospace Equipment Based on Parameter-optimized XGBoost

• Zhigang Sun
• Guotao Wang
• Guofu Zhai
• Pengfei Li
• Qi Liang
• Min Zhang

The complex internal structure and heterogeneous composition material of aerospace equipment make it difficult to apply traditional acoustic emission source localization methods to the loose particle localization research. In previous studies, the authors transformed the loose particle localization problem into the multi-classification problem in machine learning, trained the loose particle localization model, but there were problems of low classification accuracy and low practicality. In this paper, the authors first introduced the ensemble learning idea into the loose particle detection field, designed a complete loose particle localization scheme, analyzed the above problems from multiple perspectives, and proposed new algorithms or strategies to enhance the superiority and practicability of the loose particle localization model. Specifically, in view of the low classification accuracy, the authors carried out the research on pulse preprocessing, feature engineering and model training from the four perspectives of signal, feature, data set and classifier, respectively. The zero-pulse-filling pulse matching algorithm and the channel-weighting-based feature selection method was newly proposed, Mel-Frequency Cepstral Coefficients features was newly extracted, feature optimization scheme was designed, and XGBoost ensemble classifier was trained. Thus, high-quality loose particle signals, high-quality localization data sets and high-performance loose particle localization models were obtained, respectively. Test results show that, the classification accuracy achieved by the new loose particle localization model was 96. 80%, which was a significant improvement compared to the 83. 53% achieved in previous studies. In view of the low practicality, the authors built the loose particle localization experimental system, gave the specific implementation steps of the loose particle localization method, as well as the general procedures for applying the loose particle localization model for physical testing. Meanwhile, taking into account the requirements of aerospace engineering applications, the authors added the majority voting strategy to convert classification results into localization results, thus newly proposed the definition of equipment-level loose particle localization accuracy. Several physical testing results show that, the loose particle localization model achieved a localization accuracy of 90. 91%, which effectively verified the feasibility and stability of the proposed method. This study is an important supplement to the loose particle detection research, and of great significance in improving the reliability of aerospace systems.

IJCAI Conference 2023 Conference Paper

PPAT: Progressive Graph Pairwise Attention Network for Event Causality Identification

• Zhenyu Liu
• Baotian Hu
• Zhenran Xu
• Min Zhang

Event Causality Identification (ECI) aims to identify the causality between a pair of event mentions in a document, which is composed of sentence-level ECI (SECI) and document-level ECI (DECI). Previous work applies various reasoning models to identify the implicit event causality. However, they indiscriminately reason all event causality in the same way, ignoring that most inter-sentence event causality depends on intra-sentence event causality to infer. In this paper, we propose a progressive graph pairwise attention network (PPAT) to consider the above dependence. PPAT applies a progressive reasoning strategy, as it first predicts the intra-sentence event causality, and then infers the more implicit inter-sentence event causality based on the SECI result. We construct a sentence boundary event relational graph, and PPAT leverages a simple pairwise attention mechanism, which attends to different reasoning chains on the graph. In addition, we propose a causality-guided training strategy for assisting PPAT in learning causality-related representations on every layer. Extensive experiments show that our model achieves state-of-the-art performance on three benchmark datasets (5. 5%, 2. 2% and 4. 5% F1 gains on EventStoryLine, MAVEN-ERE and Causal-TimeBank). Code is available at https: //github. com/HITsz-TMG/PPAT.

TCS Journal 2023 Journal Article

Preface for the special issue of Theoretical Computer Science in honor of the 60th birthday of Yuxi Fu

• Yijia Chen
• Pierre-Louis Curien
• Yuxin Deng
• Min Zhang

IJCAI Conference 2023 Conference Paper

RaSa: Relation and Sensitivity Aware Representation Learning for Text-based Person Search

• Yang Bai
• Min Cao
• Daming Gao
• Ziqiang Cao
• Chen Chen
• Zhenfeng Fan
• Liqiang Nie
• Min Zhang

Text-based person search aims to retrieve the specified person images given a textual description. The key to tackling such a challenging task is to learn powerful multi-modal representations. Towards this, we propose a Relation and Sensitivity aware representation learning method (RaSa), including two novel tasks: Relation-Aware learning (RA) and Sensitivity-Aware learning (SA). For one thing, existing methods cluster representations of all positive pairs without distinction and overlook the noise problem caused by the weak positive pairs where the text and the paired image have noise correspondences, thus leading to overfitting learning. RA offsets the overfitting risk by introducing a novel positive relation detection task (i. e. , learning to distinguish strong and weak positive pairs). For another thing, learning invariant representation under data augmentation (i. e. , being insensitive to some transformations) is a general practice for improving representation's robustness in existing methods. Beyond that, we encourage the representation to perceive the sensitive transformation by SA (i. e. , learning to detect the replaced words), thus promoting the representation's robustness. Experiments demonstrate that RaSa outperforms existing state-of-the-art methods by 6. 94%, 4. 45% and 15. 35% in terms of Rank@1 on CUHK-PEDES, ICFG-PEDES and RSTPReid datasets, respectively. Code is available at: https: //github. com/Flame-Chasers/RaSa.

AAAI Conference 2023 Conference Paper

RenewNAT: Renewing Potential Translation for Non-autoregressive Transformer

• Pei Guo
• Yisheng Xiao
• Juntao Li
• Min Zhang

Non-autoregressive neural machine translation (NAT) models are proposed to accelerate the inference process while maintaining relatively high performance. However, existing NAT models are difficult to achieve the desired efficiency-quality trade-off. For one thing, fully NAT models with efficient inference perform inferior to their autoregressive counterparts. For another, iterative NAT models can, though, achieve comparable performance while diminishing the advantage of speed. In this paper, we propose RenewNAT, a flexible framework with high efficiency and effectiveness, to incorporate the merits of fully and iterative NAT models. RenewNAT first generates the potential translation results and then renews them in a single pass. It can achieve significant performance improvements at the same expense as traditional NAT models (without introducing additional model parameters and decoding latency). Experimental results on various translation benchmarks (e.g., 4 WMT) show that our framework consistently improves the performance of strong fully NAT methods (e.g., GLAT and DSLP) without additional speed overhead.

AAAI Conference 2023 Conference Paper

Towards Reliable Neural Machine Translation with Consistency-Aware Meta-Learning

• Rongxiang Weng
• Qiang Wang
• Wensen Cheng
• Changfeng Zhu
• Min Zhang

Neural machine translation (NMT) has achieved remarkable success in producing high-quality translations. However, current NMT systems suffer from a lack of reliability, as their outputs that are often affected by lexical or syntactic changes in inputs, resulting in large variations in quality. This limitation hinders the practicality and trustworthiness of NMT. A contributing factor to this problem is that NMT models trained with the one-to-one paradigm struggle to handle the source diversity phenomenon, where inputs with the same meaning can be expressed differently. In this work, we treat this problem as a bilevel optimization problem and present a consistency-aware meta-learning (CAML) framework derived from the model-agnostic meta-learning (MAML) algorithm to address it. Specifically, the NMT model with CAML (named CoNMT) first learns a consistent meta representation of semantically equivalent sentences in the outer loop. Subsequently, a mapping from the meta representation to the output sentence is learned in the inner loop, allowing the NMT model to translate semantically equivalent sentences to the same target sentence. We conduct experiments on the NIST Chinese to English task, three WMT translation tasks, and the TED M2O task. The results demonstrate that CoNMT effectively improves overall translation quality and reliably handles diverse inputs.

AAAI Conference 2022 Conference Paper

A Label Dependence-Aware Sequence Generation Model for Multi-Level Implicit Discourse Relation Recognition

• Changxing Wu
• Liuwen Cao
• Yubin Ge
• Yang Liu
• Min Zhang
• Jinsong Su

Implicit discourse relation recognition (IDRR) is a challenging but crucial task in discourse analysis. Most existing methods train multiple models to predict multi-level labels independently, while ignoring the dependence between hierarchically structured labels. In this paper, we consider multi-level IDRR as a conditional label sequence generation task and propose a Label Dependence-aware Sequence Generation Model (LDSGM) for it. Specifically, we first design a label attentive encoder to learn the global representation of an input instance and its level-specific contexts, where the label dependence is integrated to obtain better label embeddings. Then, we employ a label sequence decoder to output the predicted labels in a top-down manner, where the predicted higherlevel labels are directly used to guide the label prediction at the current level. We further develop a mutual learning enhanced training method to exploit the label dependence in a bottom-up direction, which is captured by an auxiliary decoder introduced during training. Experimental results on the PDTB dataset show that our model achieves the state-of-theart performance on multi-level IDRR. We release our code at https: //github. com/nlpersECJTU/LDSGM.

AAAI Conference 2022 Conference Paper

Deep Fusing Pre-trained Models into Neural Machine Translation

• Rongxiang Weng
• Heng Yu
• Weihua Luo
• Min Zhang

Pre-training and fine-tuning have become the de facto paradigm in many natural language processing (NLP) tasks. However, compared to other NLP tasks, neural machine translation (NMT) aims to generate target language sentences through the contextual representation from the source language counterparts. This characteristic means the optimization objective of NMT is far from that of the universal pre-trained models (PTMs), leading to the standard procedure of pretraining and fine-tuning does not work well in NMT. In this paper, we propose a novel framework to deep fuse the pretrained representation into NMT, fully exploring the potential of PTMs in NMT. Specifically, we directly replace the randomly initialized Transformer encoder with a pre-trained encoder and propose a layer-wise coordination structure to coordinate PTM and NMT decoder learning. Then, we introduce a partitioned multi-task learning method to fine-tune the pretrained parameter, reducing the gap between PTM and NMT by progressively learning the task-specific representation. Experimental results show that our approach achieves considerable improvements on WMT14 En2De, WMT14 En2Fr, and WMT16 Ro2En translation benchmarks and outperforms previous work in both autoregressive and non-autoregressive NMT models.

EAAI Journal 2022 Journal Article

Deep subdomain generalisation network for health monitoring of high-speed train brake pads

• Ruohui Hu
• Min Zhang
• Xiangyin Meng
• Zhuang Kang

Uneven wear at different locations in high-speed train brake pads creates inconsistent vibrations, which make the criteria learned by the intelligent model hard to standardise to determine friction block failure. Variable friction-induced vibration makes brake pad health monitoring a cross-domain diagnostic problem. This study proposes a deep subdomain generalisation network for online monitoring of the health status of train brake pads. In contrast to existing transfer learning methods, the proposed approach does not assume the availability of test data during training. The model is generalised using unsupervised learning to new scenarios with unknown specific locations of uneven wear. The network reduces the discrepancy explicitly among the distributions of relevant source subdomains using local maximum mean discrepancy, which helps subdomains with identical labels more accurately find an appropriate common subspace. Furthermore, Bayesian optimisation is used to unify feature extraction, domain generalisation and hyper-parameter optimisation into a framework. The vibration–acceleration signals of different friction blocks are collected, and generalisation experiments are performed on both braking-friction and rotating-machinery datasets. The results indicate that the proposed model outperforms the other domain generalisation approaches and accurately identifies the state of a target friction block.

IJCAI Conference 2022 Conference Paper

Effective Graph Context Representation for Document-level Machine Translation

• Kehai Chen
• Muyun Yang
• Masao Utiyama
• Eiichiro Sumita
• Rui Wang
• Min Zhang

Document-level neural machine translation (DocNMT) universally encodes several local sentences or the entire document. Thus, DocNMT does not consider the relevance of document-level contextual information, for example, some context (i. e. , content words, logical order, and co-occurrence relation) is more effective than another auxiliary context (i. e. , functional and auxiliary words). To address this issue, we first utilize the word frequency information to recognize content words in the input document, and then use heuristical relations to summarize content words and sentences as a graph structure without relying on external syntactic knowledge. Furthermore, we apply graph attention networks to this graph structure to learn its feature representation, which allows DocNMT to more effectively capture the document-level context. Experimental results on several widely-used document-level benchmarks demonstrated the effectiveness of the proposed approach.

IJCAI Conference 2022 Conference Paper

Efficient Document-level Event Extraction via Pseudo-Trigger-aware Pruned Complete Graph

• Tong Zhu
• Xiaoye Qu
• Wenliang Chen
• Zhefeng Wang
• Baoxing Huai
• Nicholas Yuan
• Min Zhang

Most previous studies of document-level event extraction mainly focus on building argument chains in an autoregressive way, which achieves a certain success but is inefficient in both training and inference. In contrast to the previous studies, we propose a fast and lightweight model named as PTPCG. In our model, we design a novel strategy for event argument combination together with a non-autoregressive decoding algorithm via pruned complete graphs, which are constructed under the guidance of the automatically selected pseudo triggers. Compared to the previous systems, our system achieves competitive results with 19. 8% of parameters and much lower resource consumption, taking only 3. 8% GPU hours for training and up to 8. 5 times faster for inference. Besides, our model shows superior compatibility for the datasets with (or without) triggers and the pseudo triggers can be the supplements for annotated triggers to make further improvements. Codes are available at https: //github. com/Spico197/DocEE.

IJCAI Conference 2022 Conference Paper

Image-text Retrieval: A Survey on Recent Research and Development

• Min Cao
• Shiping Li
• Juntao Li
• Liqiang Nie
• Min Zhang

In the past few years, cross-modal image-text retrieval (ITR) has experienced increased interest in the research community due to its excellent research value and broad real-world application. It is designed for the scenarios where the queries are from one modality and the retrieval galleries from another modality. This paper presents a comprehensive and up-to-date survey on the ITR approaches from four perspectives. By dissecting an ITR system into two processes: feature extraction and feature alignment, we summarize the recent advance of the ITR approaches from these two perspectives. On top of this, the efficiency-focused study on the ITR system is introduced as the third perspective. To keep pace with the times, we also provide a pioneering overview of the cross-modal pre-training ITR approaches as the fourth perspective. Finally, we outline the common benchmark datasets and evaluation metric for ITR, and conduct the accuracy comparison among the representative ITR approaches. Some critical yet less studied issues are discussed at the end of the paper.

NeurIPS Conference 2022 Conference Paper

LasUIE: Unifying Information Extraction with Latent Adaptive Structure-aware Generative Language Model

• Hao Fei
• Shengqiong Wu
• Jingye Li
• Bobo Li
• Fei Li
• Libo Qin
• Meishan Zhang
• Min Zhang

Universally modeling all typical information extraction tasks (UIE) with one generative language model (GLM) has revealed great potential by the latest study, where various IE predictions are unified into a linearized hierarchical expression under a GLM. Syntactic structure information, a type of effective feature which has been extensively utilized in IE community, should also be beneficial to UIE. In this work, we propose a novel structure-aware GLM, fully unleashing the power of syntactic knowledge for UIE. A heterogeneous structure inductor is explored to unsupervisedly induce rich heterogeneous structural representations by post-training an existing GLM. In particular, a structural broadcaster is devised to compact various latent trees into explicit high-order forests, helping to guide a better generation during decoding. We finally introduce a task-oriented structure fine-tuning mechanism, further adjusting the learned structures to most coincide with the end-task's need. Over 12 IE benchmarks across 7 tasks our system shows significant improvements over the baseline UIE system. Further in-depth analyses show that our GLM learns rich task-adaptive structural bias that greatly resolves the UIE crux, the long-range dependence issue and boundary identifying.

IJCAI Conference 2022 Conference Paper

Learning Discrete Representations via Constrained Clustering for Effective and Efficient Dense Retrieval (Extended Abstract)

• Jingtao Zhan
• Jiaxin Mao
• Yiqun Liu
• Jiafeng Guo
• Min Zhang
• Shaoping Ma

Dense Retrieval (DR) has achieved state-of-the-art first-stage ranking effectiveness. However, the efficiency of most existing DR models is limited by the large memory cost of storing dense vectors and the time-consuming nearest neighbor search (NNS) in vector space. Therefore, we present RepCONC, a novel retrieval model that learns discrete Representations via CONstrained Clustering. RepCONC jointly trains dual-encoders and the Product Quantization (PQ) method to learn discrete document representations and enables fast approximate NNS with compact indexes. It models quantization as a constrained clustering process, which requires the document embeddings to be uniformly clustered around the quantization centroids. We theoretically demonstrate the importance of the uniform clustering constraint and derive an efficient approximate solution for constrained clustering by reducing it to an instance of the optimal transport problem. Extensive experiments on two popular ad-hoc retrieval benchmarks show that RepCONC substantially outperforms a wide range of existing retrieval models in terms of retrieval effectiveness, memory efficiency, and time efficiency.

IJCAI Conference 2021 Conference Paper

A Structure Self-Aware Model for Discourse Parsing on Multi-Party Dialogues

• Ante Wang
• Linfeng Song
• Hui Jiang
• Shaopeng Lai
• Junfeng Yao
• Min Zhang
• Jinsong Su

Conversational discourse structures aim to describe how a dialogue is organized, thus they are helpful for dialogue understanding and response generation. This paper focuses on predicting discourse dependency structures for multi-party dialogues. Previous work adopts incremental methods that take the features from the already predicted discourse relations to help generate the next one. Although the inter-correlations among predictions considered, we find that the error propagation is also very serious and hurts the overall performance. To alleviate error propagation, we propose a Structure Self-Aware (SSA) model, which adopts a novel edge-centric Graph Neural Network (GNN) to update the information between each Elementary Discourse Unit (EDU) pair layer by layer, so that expressive representations can be learned without historical predictions. In addition, we take auxiliary training signals (e. g. structure distillation) for better representation learning. Our model achieves the new state-of-the-art performances on two conversational discourse parsing benchmarks, largely outperforming the previous methods.

AAAI Conference 2021 Conference Paper

Attributes-Guided and Pure-Visual Attention Alignment for Few-Shot Recognition

• Siteng Huang
• Min Zhang
• Yachen Kang
• Donglin Wang

The purpose of few-shot recognition is to recognize novel categories with a limited number of labeled examples in each class. To encourage learning from a supplementary view, recent approaches have introduced auxiliary semantic modalities into effective metric-learning frameworks that aim to learn a feature similarity between training samples (support set) and test samples (query set). However, these approaches only augment the representations of samples with available semantics while ignoring the query set, which loses the potential for the improvement and may lead to a shift between the modalities combination and the pure-visual representation. In this paper, we devise an attributes-guided attention module (AGAM) to utilize human-annotated attributes and learn more discriminative features. This plug-and-play module enables visual contents and corresponding attributes to collectively focus on important channels and regions for the support set. And the feature selection is also achieved for query set with only visual information while the attributes are not available. Therefore, representations from both sets are improved in a fine-grained manner. Moreover, an attention alignment mechanism is proposed to distill knowledge from the guidance of attributes to the pure-visual branch for samples without attributes. Extensive experiments and analysis show that our proposed module can significantly improve simple metric-based approaches to achieve state-of-the-art performance on different datasets and settings.

JBHI Journal 2021 Journal Article

Brain Network Analysis by Stable and Unstable EEG Components

• Shengnan Liu
• Min Li
• Yukun Feng
• Min Zhang
• Mirabel Ewura Esi Acquah
• Sunpei Huang
• Jinying Chen
• Peng Ren

Objective: Previous studies have already shown that electroencephalography (EEG) brain network (BN) can reflect the health status of individuals. However, novel methods are still needed for BN analysis. Therefore, in this study, BNs were constructed based on stable and unstable EEG components, and these may be implemented for disease diagnosis. Methods: Parkinson's disease (PD) was used as an example to illustrate this method. First, EEG signals were decomposed into dynamic modes (DMs). Each DM contains one eigenvalue that can determine not only the stability of that mode, but also its corresponding oscillatory frequency. Second, the stable and unstable components of EEG signals in each frequency band (delta, theta, alpha and beta) were calculated, which are based on the stable and unstable DMs within each respective frequency band. Third, newly developed BNs were constructed, including stable brain network (SBN), unstable brain network (UBN) and inter-connected brain network (IBN). Finally, their topological attributes were extracted in order to differentiate between PD patients and healthy controls (HC). Furthermore, topological attributes were also derived from traditional brain network (TBN) for comparison. Results: Most topological attributes of SBN, UBN and IBN can significantly differentiate between PD patients and HC (p value <; 0. 05). Furthermore, the area under the curve (AUC), precision and recall values of SBN analysis are all significantly higher than TBN. Conclusion: We proposed a new perspective on EEG BN analysis. Significance: These newly developed BNs not only have biological significance, but also could be widely applied in most medical and engineering fields.

YNIMG Journal 2021 Journal Article

Brain responses to drug cues predict craving changes in abstinent heroin users: A preliminary study

• Shuang Liu
• Shicong Wang
• Min Zhang
• Yan Xu
• Ziqiang Shao
• Longmao Chen
• Wenhan Yang
• Jun Liu

BACKGROUND: Loss of control over drug intake occurring in drug addiction is believed to result from disruption of reward circuits, including reduced responsiveness to natural rewards (e.g., monetary, sex) and heightened responsiveness to drug reward. Yet few studies have assessed reward deficiency and related brain responses in abstinent heroin users with opioid use disorder, and less is known whether the brain responses can predict cue-induced craving changes following by prolonged abstinence. METHOD: 31 heroin users (age: 44.13±7.68 years, male: 18 (58%), duration of abstinence: 85.2 ± 52.5 days) were enrolled at a mandatory detoxification center. By employing a cue-reactivity paradigm including three types of cues (drug, sexual, neutral), brain regional activations and circuit-level functional coupling were extracted. Among the 31 heroin users, 15 were followed up longitudinally to assess cue induced craving changes in the ensuing 6 months. RESULTS: One way analysis of variance results showed that heroin users have differential brain activations to the three cues (neutral, drug and sexual) in the left dorsolateral prefrontal cortex (DLPFC), insula, orbiotofrontal cortex (OFC) and the bilateral thalamus. Drug cue induced greater activations in left DLPFC, insula and OFC compared to sexual cue. The psychophysiological interactions (PPI) analysis revealed negative couplings of the left DLPFC and the left OFC, bilateral thalamus, putamen in heroin users during drug cue exposure. In the 6-month follow-up study, both drug cue induced activation of the left DLPFC and the functional coupling of the left DLPFC-bilateral thalamus at baseline was correlated with craving reductions, which were not found for sexual cues. CONCLUSION: Our preliminary study provided novel evidence for the reward deficiency theory of opioid use disorder. Our findings also have clinical implications, as drug cue induced activation of the left DLPFC and functional coupling of left DLPFC-bilateral thalamus may be potential neuroimaging markers for craving changes during prolonged abstinence. Evidently, the findings in the current preliminary study should be confirmed by large sample size in the future.

AIJ Journal 2021 Journal Article

Dependency-based syntax-aware word representations

• Meishan Zhang
• Zhenghua Li
• Guohong Fu
• Min Zhang

Dependency syntax has been demonstrated highly useful for a number of natural language processing (NLP) tasks. Typical approaches of utilizing dependency syntax include Tree-RNN and Tree-Linearization, both of which exploit explicit 1-best tree outputs from a well-trained parser as inputs. However, these approaches may suffer from error propagation due to the inevitable errors contained in the 1-best tree outputs. In this work, we propose a novel approach to integrate dependency syntax without using the discrete tree outputs. The key idea is to use the intermediate hidden representations of a well-trained encoder-decoder dependency parser, which are referred to as Dependency-based Syntax-Aware Word Representations (Dep-SAWRs). Then, we simply concatenate such Dep-SAWRs with the conventional context-insensitive word embeddings to compose input word representations, without requiring to modify the model architecture of the downstream tasks. We evaluate the proposed method on four kinds of typical NLP tasks, including sentence classification, sentence matching, sequence labeling and machine translation. Experimental results show that the proposed approach is highly promising. On the one hand, it can utilize dependency syntax effectively, bringing consistently better performance on the four tasks compared with baselines without using syntax. On the other hand, the proposed method can outperform the Tree-RNN and Tree-Linearization approaches in most settings, and meanwhile are highly efficient in syntax integration. In addition, the proposed method would be easily extendable to encoding other structural attributes of language.

AAAI Conference 2021 Conference Paper

Graph Heterogeneous Multi-Relational Recommendation

• Chong Chen
• Weizhi Ma
• Min Zhang
• Zhaowei Wang
• Xiuqiang He
• Chenyang Wang
• Yiqun Liu
• Shaoping Ma

Traditional studies on recommender systems usually leverage only one type of user behaviors (the optimization target, such as purchase), despite the fact that users also generate a large number of various types of interaction data (e. g. , view, click, add-to-cart, etc). Generally, these heterogeneous multirelational data provide well-structured information and can be used for high-quality recommendation. Early efforts towards leveraging these heterogeneous data fail to capture the high-hop structure of user-item interactions, which are unable to make full use of them and may only achieve constrained recommendation performance. In this work, we propose a new multi-relational recommendation model named Graph Heterogeneous Collaborative Filtering (GHCF). To explore the high-hop heterogeneous user-item interactions, we take the advantages of Graph Convolutional Network (GCN) and further improve it to jointly embed both representations of nodes (users and items) and relations for multi-relational prediction. Moreover, to fully utilize the whole heterogeneous data, we perform the advanced efficient non-sampling optimization under a multi-task learning framework. Experimental results on two public benchmarks show that GHCF significantly outperforms the state-of-the-art recommendation methods, especially for cold-start users who have few primary item interactions. Further analysis verifies the importance of the proposed embedding propagation for modelling high-hop heterogeneous user-item interactions, showing the rationality and effectiveness of GHCF. Our implementation has been released (https: //github. com/chenchongthu/GHCF).

IJCAI Conference 2021 Conference Paper

Improving Context-Aware Neural Machine Translation with Source-side Monolingual Documents

• Linqing Chen
• Junhui Li
• Zhengxian Gong
• Xiangyu Duan
• Boxing Chen
• Weihua Luo
• Min Zhang
• Guodong Zhou

Document context-aware machine translation remains challenging due to the lack of large-scale document parallel corpora. To make full use of source-side monolingual documents for context-aware NMT, we propose a Pre-training approach with Global Context (PGC). In particular, we first propose a novel self-supervised pre-training task, which contains two training objectives: (1) reconstructing the original sentence from a corrupted version; (2) generating a gap sentence from its left and right neighbouring sentences. Then we design a universal model for PGC which consists of a global context encoder, a sentence encoder and a decoder, with similar architecture to typical context-aware NMT models. We evaluate the effectiveness and generality of our pre-trained PGC model by adapting it to various downstream context-aware NMT models. Detailed experimentation on four different translation tasks demonstrates that our PGC approach significantly improves the translation performance of context-aware NMT. For example, based on the state-of-the-art SAN model, we achieve an averaged improvement of 1. 85 BLEU scores and 1. 59 Meteor scores on the four translation tasks.

NeurIPS Conference 2021 Conference Paper

R-Drop: Regularized Dropout for Neural Networks

• Xiaobo Liang
• Lijun Wu
• Juntao Li
• Yue Wang
• Qi Meng
• Tao Qin
• Wei Chen
• Min Zhang

Dropout is a powerful and widely used technique to regularize the training of deep neural networks. Though effective and performing well, the randomness introduced by dropout causes unnegligible inconsistency between training and inference. In this paper, we introduce a simple consistency training strategy to regularize dropout, namely R-Drop, which forces the output distributions of different sub models generated by dropout to be consistent with each other. Specifically, for each training sample, R-Drop minimizes the bidirectional KL-divergence between the output distributions of two sub models sampled by dropout. Theoretical analysis reveals that R-Drop reduces the above inconsistency. Experiments on $\bf{5}$ widely used deep learning tasks ($\bf{18}$ datasets in total), including neural machine translation, abstractive summarization, language understanding, language modeling, and image classification, show that R-Drop is universally effective. In particular, it yields substantial improvements when applied to fine-tune large-scale pre-trained models, e. g. , ViT, RoBERTa-large, and BART, and achieves state-of-the-art (SOTA) performances with the vanilla Transformer model on WMT14 English$\to$German translation ($\bf{30. 91}$ BLEU) and WMT14 English$\to$French translation ($\bf{43. 95}$ BLEU), even surpassing models trained with extra large-scale data and expert-designed advanced variants of Transformer models. Our code is available at GitHub\footnote{\url{https: //github. com/dropreg/R-Drop}}.

AAAI Conference 2021 Conference Paper

Tightening Robustness Verification of Convolutional Neural Networks with Fine-Grained Linear Approximation

• Yiting Wu
• Min Zhang

The robustness of neural networks can be quantitatively indicated by a lower bound within which any perturbation does not alter the original input’s classification result. A certified lower bound is also a criterion to evaluate the performance of robustness verification approaches. In this paper, we present a tighter linear approximation approach for the robustness verification of Convolutional Neural Networks (CNNs). By the tighter approximation, we can tighten the robustness verification of CNNs, i. e. , proving they are robust within a larger perturbation distance. Furthermore, our approach is applicable to general sigmoid-like activation functions. We implement DeepCert, the resulting verification toolkit. We evaluate it with open-source benchmarks, including LeNet and the models trained on MNIST and CIFAR. Experimental results show that DeepCert outperforms other state-of-the-art robustness verification tools with at most 286. 3% improvement to the certified lower bound and 1566. 8 times speedup for the same neural networks.

IJCAI Conference 2021 Conference Paper

Transfer Learning via Optimal Transportation for Integrative Cancer Patient Stratification

• Ziyu Liu
• WEI SHAO
• Jie Zhang
• Min Zhang
• Kun Huang

The Stratification of early-stage cancer patients for the prediction of clinical outcome is a challenging task since cancer is associated with various molecular aberrations. A single biomarker often cannot provide sufficient information to stratify early-stage patients effectively. Understanding the complex mechanism behind cancer development calls for exploiting biomarkers from multiple modalities of data such as histopathology images and genomic data. The integrative analysis of these biomarkers sheds light on cancer diagnosis, subtyping, and prognosis. Another difficulty is that labels for early-stage cancer patients are scarce and not reliable enough for predicting survival times. Given the fact that different cancer types share some commonalities, we explore if the knowledge learned from one cancer type can be utilized to improve prognosis accuracy for another cancer type. We propose a novel unsupervised multi-view transfer learning algorithm to simultaneously analyze multiple biomarkers in different cancer types. We integrate multiple views using non-negative matrix factorization and formulate the transfer learning model based on the Optimal Transport theory to align features of different cancer types. We evaluate the stratification performance on three early-stage cancers from the Cancer Genome Atlas (TCGA) project. Comparing with other benchmark methods, our framework achieves superior accuracy for patient outcome prediction.

AAAI Conference 2020 Conference Paper

Alignment-Enhanced Transformer for Constraining NMT with Pre-Specified Translations

• Kai Song
• Kun Wang
• Heng Yu
• Yue Zhang
• Zhongqiang Huang
• Weihua Luo
• Xiangyu Duan
• Min Zhang

We investigate the task of constraining NMT with prespecified translations, which has practical significance for a number of research and industrial applications. Existing works impose pre-specified translations as lexical constraints during decoding, which are based on word alignments derived from target-to-source attention weights. However, multiple recent studies have found that word alignment derived from generic attention heads in the Transformer is unreliable. We address this problem by introducing a dedicated head in the multi-head Transformer architecture to capture external supervision signals. Results on five language pairs show that our method is highly effective in constraining NMT with pre-specified translations, consistently outperforming previous methods in translation quality.

IJCAI Conference 2020 Conference Paper

BERT-PLI: Modeling Paragraph-Level Interactions for Legal Case Retrieval

• Yunqiu Shao
• Jiaxin Mao
• Yiqun Liu
• Weizhi Ma
• Ken Satoh
• Min Zhang
• Shaoping Ma

Legal case retrieval is a specialized IR task that involves retrieving supporting cases given a query case. Compared with traditional ad-hoc text retrieval, the legal case retrieval task is more challenging since the query case is much longer and more complex than common keyword queries. Besides that, the definition of relevance between a query case and a supporting case is beyond general topical relevance and it is therefore difficult to construct a large-scale case retrieval dataset, especially one with accurate relevance judgments. To address these challenges, we propose BERT-PLI, a novel model that utilizes BERT to capture the semantic relationships at the paragraph-level and then infers the relevance between two cases by aggregating paragraph-level interactions. We fine-tune the BERT model with a relatively small-scale case law entailment dataset to adapt it to the legal scenario and employ a cascade framework to reduce the computational cost. We conduct extensive experiments on the benchmark of the relevant case retrieval task in COLIEE 2019. Experimental results demonstrate that our proposed method outperforms existing solutions.

AAAI Conference 2020 Conference Paper

Cross-Lingual Pre-Training Based Transfer for Zero-Shot Neural Machine Translation

• Baijun Ji
• Zhirui Zhang
• Xiangyu Duan
• Min Zhang
• Boxing Chen
• Weihua Luo

Transfer learning between different language pairs has shown its effectiveness for Neural Machine Translation (NMT) in low-resource scenario. However, existing transfer methods involving a common target language are far from success in the extreme scenario of zero-shot translation, due to the language space mismatch problem between transferor (the parent model) and transferee (the child model) on the source side. To address this challenge, we propose an effective transfer learning approach based on cross-lingual pre-training. Our key idea is to make all source languages share the same feature space and thus enable a smooth transition for zero-shot translation. To this end, we introduce one monolingual pretraining method and two bilingual pre-training methods to obtain a universal encoder for different languages. Once the universal encoder is constructed, the parent model built on such encoder is trained with large-scale annotated data and then directly applied in zero-shot translation scenario. Experiments on two public datasets show that our approach significantly outperforms strong pivot-based baseline and various multilingual NMT approaches.

YNICL Journal 2020 Journal Article

Dysfunction of the NAc-mPFC circuit in insomnia disorder

• Ziqiang Shao
• Yan Xu
• Longmao Chen
• Shicong Wang
• Min Zhang
• Shuang Liu
• Xinwen Wen
• Dahua Yu

BACKGROUND: Insomnia disorder (ID) is a prevalent sleep disorder, which seriously affects people's daily life and was found to be associated with increased frequency of sleep stage shifts. Previous findings had revealed the critical role of the nucleus accumbens (NAc) in sleep-wake transition. However, the neuroimaging studies of the NAc in patients with ID have been rare. We hypothesized that structural and functional abnormalities of the NAc would be implicated in ID. METHODS: Twenty-six ID patients and 36 matched healthy controls (HC) were included in the current study. The volumes and corresponding resting-state functional connectivity (RSFC) of the bilateral NAc were compared between the two groups. The abnormal RSFC in ID were then correlated with Pittsburgh Sleep Quality Index (PSQI). RESULTS: Compared with HC, ID patients showed significantly increased volume of right NAc. Several brain regions showed increased RSFC with the NAc in ID patients, such as medial prefrontal cortex (mPFC), anterior cingulate cortex (ACC), caudate and putamen. Meanwhile, the occipital gyrus and temporal gyrus showed decreased RSFC with the NAc. Additionally, the increased RSFC strength between bilateral NAc and left mPFC was significant correlated with PSQI scores in ID patients. CONCLUSION: Dysfunctions of the NAc-mPFC circuit were found in ID patients, which were associated with sleep quality measured by PSQI. The two patterns of increase and decrease of RSFC in ID patients observed in our study may reflect the state of hyperarousal and potential impairment of cognitive function in the patients, respectively. It is hoped that our study focusing on NAc-mPFC circuits could provide new insights for the neural mechanisms of ID and potential novel therapeutic targets for treatment of ID patients.

AAAI Conference 2020 Conference Paper

Efficient Heterogeneous Collaborative Filtering without Negative Sampling for Recommendation

• Chong Chen
• Min Zhang
• Yongfeng Zhang
• Weizhi Ma
• Yiqun Liu
• Shaoping Ma

Recent studies on recommendation have largely focused on exploring state-of-the-art neural networks to improve the expressiveness of models, while typically apply the Negative Sampling (NS) strategy for efficient learning. Despite effectiveness, two important issues have not been well-considered in existing methods: 1) NS suffers from dramatic fluctuation, making sampling-based methods difficult to achieve the optimal ranking performance in practical applications; 2) although heterogeneous feedback (e. g. , view, click, and purchase) is widespread in many online systems, most existing methods leverage only one primary type of user feedback such as purchase. In this work, we propose a novel nonsampling transfer learning solution, named Efficient Heterogeneous Collaborative Filtering (EHCF) for Top-N recommendation. It can not only model fine-grained user-item relations, but also efficiently learn model parameters from the whole heterogeneous data (including all unlabeled data) with a rather low time complexity. Extensive experiments on three real-world datasets show that EHCF significantly outperforms state-of-the-art recommendation methods in both traditional (single-behavior) and heterogeneous scenarios. Moreover, EHCF shows significant improvements in training ef- ficiency, making it more applicable to real-world large-scale systems. Our implementation has been released 1 to facilitate further developments on efficient whole-data based neural methods.

AAAI Conference 2020 Conference Paper

Improving Neural Relation Extraction with Positive and Unlabeled Learning

• Zhengqiu He
• Wenliang Chen
• Yuyi Wang
• Wei Zhang
• Guanchun Wang
• Min Zhang

We present a novel approach to improve the performance of distant supervision relation extraction with Positive and Unlabeled (PU) Learning. This approach first applies reinforcement learning to decide whether a sentence is positive to a given relation, and then positive and unlabeled bags are constructed. In contrast to most previous studies, which mainly use selected positive instances only, we make full use of unlabeled instances and propose two new representations for positive and unlabeled bags. These two representations are then combined in an appropriate way to make bag-level prediction. Experimental results on a widely used real-world dataset demonstrate that this new approach indeed achieves significant and consistent improvements as compared to several competitive baselines.

AAAI Conference 2020 Conference Paper

Sentiment Classification in Customer Service Dialogue with Topic-Aware Multi-Task Learning

• Jiancheng Wang
• Jingjing Wang
• Changlong Sun
• Shoushan Li
• Xiaozhong Liu
• Luo Si
• Min Zhang
• Guodong Zhou

Sentiment analysis in dialogues plays a critical role in dialogue data analysis. However, previous studies on sentiment classification in dialogues largely ignore topic information, which is important for capturing overall information in some types of dialogues. In this study, we focus on the sentiment classification task in an important type of dialogue, namely customer service dialogue, and propose a novel approach which captures overall information to enhance the classification performance. Specifically, we propose a topic-aware multi-task learning (TML) approach which learns topicenriched utterance representations in customer service dialogue by capturing various kinds of topic information. In the experiment, we propose a large-scale and high-quality annotated corpus for the sentiment classification task in customer service dialogue and empirical studies on the proposed corpus show that our approach significantly outperforms several strong baselines.

IJCAI Conference 2019 Conference Paper

Profit-driven Task Assignment in Spatial Crowdsourcing

• Jinfu Xia
• Yan Zhao
• Guanfeng Liu
• Jiajie Xu
• Min Zhang
• Kai Zheng

In Spatial Crowdsourcing (SC) systems, mobile users are enabled to perform spatio-temporal tasks by physically traveling to specified locations with the SC platforms. SC platforms manage the systems and recruit mobile users to contribute to the SC systems, whose commercial success depends on the profit attained from the task requesters. In order to maximize its profit, an SC platform needs an online management mechanism to assign the tasks to suitable workers. How to assign the tasks to workers more cost-effectively with the spatio-temporal constraints is one of the most difficult problems in SC. To deal with this challenge, we propose a novel Profit-driven Task Assignment (PTA) problem, which aims to maximize the profit of the platform. Specifically, we first establish a task reward pricing model with tasks' temporal constraints (i. e. , expected completion time and deadline). Then we adopt an optimal algorithm based on tree decomposition to achieve the optimal task assignment and propose greedy algorithms to improve the computational efficiency. Finally, we conduct extensive experiments using real and synthetic datasets, verifying the practicability of our proposed methods.

IJCAI Conference 2019 Conference Paper

Self-attentive Biaffine Dependency Parsing

• Ying Li
• Zhenghua Li
• Min Zhang
• Rui Wang
• Sheng Li
• Luo Si

The current state-of-the-art dependency parsing approaches employ BiLSTMs to encode input sentences. Motivated by the success of the transformer-based machine translation, this work for the first time applies the self-attention mechanism to dependency parsing as the replacement of the BiLSTM-based encoders, leading to competitive performance on both English and Chinese benchmark data. Based on the detailed error analysis, we then combine the power of both BiLSTM and self-attention via model ensembles, demonstrating their complementary capability of capturing contextual information. Finally, we explore the recently proposed contextualized word representations as extra input features, and further improve the parsing performance.

AIJ Journal 2019 Journal Article

Syntax-aware entity representations for neural relation extraction

• Zhengqiu He
• Wenliang Chen
• Zhenghua Li
• Wei Zhang
• Hao Shao
• Min Zhang

Distantly supervised relation extraction has been widely used to find novel relational facts between entities from text, and can be easily scaled to very large corpora. Previous studies on neural relation extraction treat this task as a multi-instance learning problem, and encode the sentences in low-dimensional spaces via neural networks. Although great progress has been made, they seldom consider the information represented by entities, which are of great significance to relation extraction. In this article, we propose several methods based on different tree-based models to learn syntax-aware entity representations for neural relation extraction. First, we encode the context of entities on dependency trees as sentence-level entity embedding based on tree-structured neural network models. Then, we utilize inter-sentence attention mechanism to obtain sentence bag level entity embedding over all sentences containing the specified entity pair. Finally, we combine both sentence embedding and entity embedding for relation classification. Experimental results on a widely used real-world dataset indicate that our system performs better than the state-of-the-art systems of relation extraction.

AAAI Conference 2019 Conference Paper

Syntax-Aware Neural Semantic Role Labeling

• Qingrong Xia
• Zhenghua Li
• Min Zhang
• Meishan Zhang
• Guohong Fu
• Rui Wang
• Luo Si

Semantic role labeling (SRL), also known as shallow semantic parsing, is an important yet challenging task in NLP. Motivated by the close correlation between syntactic and semantic structures, traditional discrete-feature-based SRL approaches make heavy use of syntactic features. In contrast, deep-neural-network-based approaches usually encode the input sentence as a word sequence without considering the syntactic structures. In this work, we investigate several previous approaches for encoding syntactic trees, and make a thorough study on whether extra syntax-aware representations are beneficial for neural SRL models. Experiments on the benchmark CoNLL-2005 dataset show that syntax-aware SRL approaches can effectively improve performance over a strong baseline with external word representations from ELMo. With the extra syntax-aware representations, our approaches achieve new state-of-the-art 85. 6 F1 (single model) and 86. 6 F1 (ensemble) on the test data, outperforming the corresponding strong baselines with ELMo by 0. 8 and 1. 0, respectively. Detailed error analysis are conducted to gain more insights on the investigated approaches.

JMLR Journal 2018 Journal Article

A Two-Stage Penalized Least Squares Method for Constructing Large Systems of Structural Equations

• Chen Chen
• Min Ren
• Min Zhang
• Dabao Zhang

We propose a two-stage penalized least squares method to build large systems of structural equations based on the instrumental variables view of the classical two-stage least squares method. We show that, with large numbers of endogenous and exogenous variables, the system can be constructed via consistent estimation of a set of conditional expectations at the first stage, and consistent selection of regulatory effects at the second stage. While the consistent estimation at the first stage can be obtained via the ridge regression, the adaptive lasso is employed at the second stage to achieve the consistent selection. This method is computationally fast and allows for parallel implementation. We demonstrate its effectiveness via simulation studies and real data analysis. [abs] [ pdf ][ bib ] &copy JMLR 2018. ( edit, beta )

AAAI Conference 2018 Conference Paper

Adversarial Learning for Chinese NER From Crowd Annotations

• YaoSheng Yang
• Meishan Zhang
• Wenliang Chen
• Wei Zhang
• Haofen Wang
• Min Zhang

To quickly obtain new labeled data, we can choose crowdsourcing as an alternative way at lower cost in a short time. But as an exchange, crowd annotations from non-experts may be of lower quality than those from experts. In this paper, we propose an approach to performing crowd annotation learning for Chinese Named Entity Recognition (NER) to make full use of the noisy sequence labels from multiple annotators. Inspired by adversarial learning, our approach uses a common Bi-LSTM and a private Bi-LSTM for representing annotatorgeneric and -specific information. The annotator-generic information is the common knowledge for entities easily mastered by the crowd. Finally, we build our Chinese NE tagger based on the LSTM-CRF model. In our experiments, we create two data sets for Chinese NER tasks from two domains. The experimental results show that our system achieves better scores than strong baseline systems.

IJCAI Conference 2018 Conference Paper

Aspect Sentiment Classification with both Word-level and Clause-level Attention Networks

• Jingjing Wang
• Jie Li
• Shoushan Li
• Yangyang Kang
• Min Zhang
• Luo Si
• Guodong Zhou

Aspect sentiment classification, a challenging task in sentiment analysis, has been attracting more and more attention in recent years. In this paper, we highlight the need for incorporating the importance degrees of both words and clauses inside a sentence and propose a hierarchical network with both word-level and clause-level attentions to aspect sentiment classification. Specifically, we first adopt sentence-level discourse segmentation to segment a sentence into several clauses. Then, we leverage multiple Bi-directional LSTM layers to encode all clauses and propose a word-level attention layer to capture the importance degrees of words in each clause. Third and finally, we leverage another Bi-directional LSTM layer to encode the outputs from the former layers and propose a clause-level attention layer to capture the importance degrees of all the clauses inside a sentence. Experimental results on the laptop and restaurant datasets from SemEval-2015 demonstrate the effectiveness of our proposed approach to aspect sentiment classification.

TCS Journal 2018 Journal Article

From hidden to visible: A unified framework for transforming behavioral theories into rewrite theories

• Min Zhang
• Kazuhiro Ogata

Algebraic formalization and verification are effective and practical ways of modeling and verifying software systems by both model checking and theorem proving techniques. In algebraic approaches, a system can be modeled either in a hidden way as a behavioral theory or in a visible way as a rewrite theory. Several approaches have been proposed to transform behavioral theories into rewrite theories for integrating model checking and theorem proving in verification. In this paper, we propose a framework for transforming behavioral theories into rewrite theories, which unifies four existing related transformation approaches. In this framework, each existing transformation approach can be viewed as a process of transforming behavioral theories first into a special class of behavioral theories and finally into rewrite theories. From this perspective, these transformation approaches differ from each other only in the transformation from ordinary behavioral theories into the classified ones, and their transformations from the classified ones into rewrite theories are essentially the same. We prove that the transformation framework preserves linear-time properties. The preservation of linear-time properties guarantees that a counterexample found by model checking a linear-time property with a generated rewrite theory is also a counterexample in the original behavioral theory, as required by integrated verification.

AAAI Conference 2018 Conference Paper

Improved English to Russian Translation by Neural Suffix Prediction

• Kai Song
• Yue Zhang
• Min Zhang
• Weihua Luo

Neural machine translation (NMT) suffers a performance de- ficiency when a limited vocabulary fails to cover the source or target side adequately, which happens frequently when dealing with morphologically rich languages. To address this problem, previous work focused on adjusting translation granularity or expanding the vocabulary size. However, morphological information is relatively under-considered in NMT architectures, which may further improve translation quality. We propose a novel method, which can not only reduce data sparsity but also model morphology through a simple but effective mechanism. By predicting the stem and suf- fix separately during decoding, our system achieves an improvement of up to 1. 98 BLEU compared with previous work on English to Russian translation. Our method is orthogonal to different NMT architectures and stably gains improvements on various domains.

AAAI Conference 2018 Conference Paper

SEE: Syntax-Aware Entity Embedding for Neural Relation Extraction

• Zhengqiu He
• Wenliang Chen
• Zhenghua Li
• Meishan Zhang
• Wei Zhang
• Min Zhang

Distant supervised relation extraction is an efficient approach to scale relation extraction to very large corpora, and has been widely used to find novel relational facts from plain text. Recent studies on neural relation extraction have shown great progress on this task via modeling the sentences in lowdimensional spaces, but seldom considered syntax information to model the entities. In this paper, we propose to learn syntax-aware entity embedding for neural relation extraction. First, we encode the context of entities on a dependency tree as sentence-level entity embedding based on tree-GRU. Then, we utilize both intra-sentence and inter-sentence attentions to obtain sentence set-level entity embedding over all sentences containing the focus entity pair. Finally, we combine both sentence embedding and entity embedding for relation classi- fication. We conduct experiments on a widely used real-world dataset and the experimental results show that our model can make full use of all informative instances and achieve stateof-the-art performance of relation extraction.

IJCAI Conference 2018 Conference Paper

Your Tweets Reveal What You Like: Introducing Cross-media Content Information into Multi-domain Recommendation

• Weizhi Ma
• Min Zhang
• Chenyang Wang
• Cheng Luo
• Yiqun Liu
• Shaoping Ma

Cold start is a challenging problem in recommender systems. Many previous studies attempt to utilize extra information from other platforms to alleviate the problem. Most of the leveraged information is on-topic, directly related to users' preferences in the target domain. Thought to be unrelated, users' off-topic content information (such as user tweets) is usually omitted. However, the off-topic content information also helps to indicate the similarity of users on their tastes, interests, and opinions, which matches the underlying assumption of Collaborative Filtering (CF) algorithms. In this paper, we propose a framework to capture the features from user's off-topic content information in social media and introduce them into Matrix Factorization (MF) based algorithms. The framework is easy to understand and flexible in different embedding approaches and MF based algorithms. To the best of our knowledge, there is no previous study in which user's off-topic content in other platforms is taken into consideration. By capturing the cross-platform content including both on-topic and off-topic information, multiple algorithms with several embedding learning approaches have achieved significant improvements in rating prediction on three datasets. Especially in cold start scenarios, we observe greater enhancement. The results confirm our suggestion that off-topic cross-media information also contributes to the recommendation.

AAAI Conference 2017 Conference Paper

Neural Machine Translation Advised by Statistical Machine Translation

• Xing Wang
• Zhengdong Lu
• Zhaopeng Tu
• Hang Li
• Deyi Xiong
• Min Zhang

Neural Machine Translation (NMT) is a new approach to machine translation that has made great progress in recent years. However, recent studies show that NMT generally produces fluent but inadequate translations (Tu et al. 2016b; 2016a; He et al. 2016; Tu et al. 2017). This is in contrast to conventional Statistical Machine Translation (SMT), which usually yields adequate but non-fluent translations. It is natural, therefore, to leverage the advantages of both models for better translations, and in this work we propose to incorporate SMT model into NMT framework. More specifically, at each decoding step, SMT offers additional recommendations of generated words based on the decoding information from NMT (e. g. , the generated partial translation and attention history). Then we employ an auxiliary classifier to score the SMT recommendations and a gating function to combine the SMT recommendations with NMT generations, both of which are jointly trained within the NMT architecture in an end-to-end manner. Experimental results on Chinese-English translation show that the proposed approach achieves significant and consistent improvements over state-of-the-art NMT and SMT systems on multiple NIST test sets.

AIJ Journal 2016 Journal Article

Exploiting meta features for dependency parsing and part-of-speech tagging

• Wenliang Chen
• Min Zhang
• Yue Zhang
• Xiangyu Duan

In recent years, discriminative methods have achieved much progress in natural language processing tasks, such as parsing, part-of-speech tagging, and word segmentation. For these methods, conventional features in a relatively high dimensional feature space may suffer from sparseness and thus exhibit less discriminative power on unseen data. This article presents a learning framework of feature transformation, addressing the sparseness problem by transforming sparse conventional base features into less sparse high-level features (i. e. meta features) with the help of a large amount of automatically annotated data. The meta features are derived by bucketing similar base features according to the frequency in large data, and used together with base features in our final system. We apply the framework to part-of-speech tagging and dependency parsing. Experimental results show that our systems perform better than the baseline systems in both tasks on standard evaluation. For the dependency parsing task, our parsers achieve state-of-the-art accuracy on the Chinese data and comparable accuracy with the best known systems on the English data. Further analysis indicates that our proposed approach is effective in processing unseen data and features.

IJCAI Conference 2016 Conference Paper

Multi-Grained Role Labeling Based on Multi-Modality Information for Real Customer Service Telephone Conversation

• Weizhi Ma
• Min Zhang
• Yiqun Liu
• Shaoping Ma

Large-scale customer service call records include lots of valuable information for business intelligence. However, the analysis of those records has not utilized in the big data era before. There are two fundamental problems before mining and analyses: 1) The telephone conversation is mixed with words of agents and users which have to be recognized before analysis; 2) The speakers in conversation are not in a pre-defined set. These problems are new challenges which have not been well studied in the previous work. In this paper, we propose a four-phase framework for role labeling in real customer service telephone conversation, with the benefit of integrating multi-modality features, i. e. , both low-level acoustic features and semantic-level textual features. Firstly, we conduct &Delta; Bayesian Information Criterion (&Delta; BIC) based speaker diarization to get two segments clusters from an audio stream. Secondly, the segments are transferred into text in an Automatic Speech Recognition (ASR) phase with a deep learning model DNN-HMM. Thirdly, by integrating acoustic and textual features, dialog level role labeling is proposed to map the two clusters into the agent and the user. Finally, sentence level role correction is designed in order to label results correctly in a fine-grained notion, which reduces the errors made in previous phases. The proposed framework is tested on two real datasets: mobile and bank customer service calls datasets. The precision of dialog level labeling is over 99. 0%. On the sentence level, the accuracy of labeling reaches 90. 4%, greatly outperforming traditional acoustic features based method which achieves only 78. 5% in accuracy.

IJCAI Conference 2016 Conference Paper

Pay Me and I'll Follow You: Detection of Crowdturfing Following Activities in Microblog Environment

• Yuli Liu
• Yiqun Liu
• Min Zhang
• Shaoping Ma

A number of existing works have focused on the problem of malicious following activity detection in microblog services. However, most of them make the assumption that the spamming following relationships are either from fraudulent accounts or compromised legitimate users. They therefore developed detection methodologies based on the features derived from this assumption. Recently, a new type of malicious crowdturfing following relationship is provided by the follower market, called voluntary following. Followers who provide voluntary following services (or named volowers) are normal users who are willing to trade their following activities for profit. Since most of their behaviors follow normal patterns, it is difficult for existing methods to detect volowers and their corresponding customers. In this work, we try to solve the voluntary following problem through a newly proposed detection method named DetectVC. This method incorporates both structure information in user following behavior graphs and prior knowledge collected from follower markets. Experimental results on large scale practical microblog data set show that DetectVC is able to detect volowers and their customers simultaneously and it also significantly outperforms existing solutions.

IJCAI Conference 2016 Conference Paper

Rating-Boosted Latent Topics: Understanding Users and Items with Ratings and Reviews

• Yunzhi Tan
• Min Zhang
• Yiqun Liu
• Shaoping Ma

The performance of a recommendation system relies heavily on the feedback of users. Most of the traditional recommendation algorithms based only on historical ratings will encounter several difficulties given the problem of data sparsity. Users' feedback usually contains rich textual reviews in addition to numerical ratings. In this paper, we exploit textual review information, as well as ratings, to model user preferences and item features in a shared topic space and subsequently introduce them into a matrix factorization model for recommendation. To this end, the data sparsity problem is alleviated and good interpretability of the recommendation results is gained. Another contribution of this work is that we model the item feature distributions with rating-boosted reviews which combine textual reviews with user sentiments. Experimental results on 26 real-world datasets from Amazon demonstrate that our approach significantly improves the rating prediction accuracy compared with various state-of-the-art models, such as LFM, HFT, CTR and RMR models. And much higher improvement is achieved for users who have few ratings, which verifies the effectiveness of the proposed approach for sparse data. Moreover, our method also benefits much from reviews on top-N recommendation tasks.

IJCAI Conference 2015 Conference Paper

Catch the Black Sheep: Unified Framework for Shilling Attack Detection Based on Fraudulent Action Propagation

• Yongfeng Zhang
• Yunzhi Tan
• Min Zhang
• Yiqun Liu
• Tat-Seng Chua
• Shaoping Ma

Many e-commerce systems allow users to express their opinions towards products through user reviews systems. The user generated reviews not only help other users to gain a more insightful view of the products, but also help online businesses to make targeted improvements on the products or services. Besides, they compose the key component of various personalized recommender systems. However, the existence of spam user accounts in the review systems introduce unfavourable disturbances into personalized recommendation by promoting or degrading targeted items intentionally through fraudulent reviews. Previous shilling attack detection algorithms usually deal with a specific kind of attacking strategy, and are exhausted to handle with the continuously emerging new cheating methods. In this work, we propose to conduct shilling attack detection for more informed recommendation by fraudulent action propagation on the reviews themselves, without caring about the specific underlying cheating strategy, which allows us a unified and flexible framework to detect the spam users.

IJCAI Conference 2015 Conference Paper

Detecting Promotion Campaigns in Community Question Answering

• Xin Li
• Yiqun Liu
• Min Zhang
• Shaoping Ma
• Xuan Zhu
• Jiashen Sun

With Community Question Answering (CQA) evolving into a quite popular method for information seeking and providing, it also becomes a target for spammers to disseminate promotion campaigns. Although there are a number of quality estimation efforts on the CQA platform, most of these works focus on identifying and reducing lowquality answers, which are mostly generated by impatient or inexperienced answerers. However, a large number of promotion answers appear to provide high-quality information to cheat CQA users in future interactions. Therefore, most existing quality estimation works in CQA may fail to detect these specially designed answers or question-answer pairs. In contrast to these works, we focus on the promotion channels of spammers, which include (shortened) URLs, telephone numbers and social media accounts. Spammers rely on these channels to connect to users to achieve promotion goals so they are irreplaceable for spamming activities. We propose a propagation algorithm to diffuse promotion intents on an “answerer-channel” bipartite graph and detect possible spamming activities. A supervised learning framework is also proposed to identify whether a QA pair is spam based on propagated promotion intents. Experimental results based on more than 6 million entries from a popular Chinese CQA portal show that our approach outperforms a number of existing quality estimation methods for detecting promotion campaigns on both the answer level and QA pair level.

AAAI Conference 2014 Conference Paper

Fraudulent Support Telephone Number Identification Based on Co-Occurrence Information on the Web

• Xin Li
• Yiqun Liu
• Min Zhang
• Shaoping Ma

“Fraudulent support phones” refers to the misleading telephone numbers placed on Web pages or other media that claim to provide services with which they are not associated. Most fraudulent support phone information is found on search engine result pages (SERPs), and such information substantially degrades the search engine user experience. In this paper, we propose an approach to identify fraudulent support telephone numbers on the Web based on the co-occurrence relations between telephone numbers that appear on SERPs. We start from a small set of seed official support phone numbers and seed fraudulent numbers. Then, we construct a co-occurrence graph according to the co-occurrence relationships of the telephone numbers that appear on Web pages. Additionally, we take the page layout information into consideration on the assumption that telephone numbers that appear in nearby page blocks should be regarded as more closely related. Finally, we develop a propagation algorithm to diffuse the trust scores of seed official support phone numbers and the distrust scores of the seed fraudulent numbers on the co-occurrence graph to detect additional fraudulent numbers. Experimental results based on over 1. 5 million SERPs produced by a popular Chinese commercial search engine indicate that our approach outperforms TrustRank, Anti-TrustRank and Good-Bad Rank algorithms by achieving an AUC value of over 0. 90.

JBHI Journal 2014 Journal Article

The Assistant Function of Three-Dimensional Information for I$^{\bf 125}$ Particle Implantation

• Wenbo Wu
• Jin Xue
• Ping Liang
• Zhigang Cheng
• Min Zhang
• Mengjuan Mu
• Cai Qi

The purpose of this study was to explore the assistant function of 3-D information for I 125 particle implantation of multineedle intervention under the guidance of ultrasound. The assistant function of 3-D information was verified by a simulation experiment system which consists of an ultrasound probe, an abdominal phantom, the preoperative computed tomography image of a patient, the electromagnetic tracking device, and the self-developed 3-D image navigation software with a practical and friendly graphical user interface. The simulation particle implantation experiments were divided into the two groups. The first group of experiments was performed with the aid of 3-D information. Seven days later, the second group of experiments was carried out with the aid of 2-D information. We made the statistical analysis of the experimental results obtained by nine medical students, nine interventional radiologists, and nine attending physicians. With the assistance of 3-D information, the percentage of tumor coverage increased (p <; 0. 01), the operation time shortened (p <; 0. 01), and the number of insertions reduced (p <; 0. 01). The assistant function of 3-D information for particle implantation of multineedle intervention under the guidance of ultrasound was technically feasible and effective.

AAAI Conference 2013 Conference Paper

A Topic-Based Coherence Model for Statistical Machine Translation

• Deyi Xiong
• Min Zhang

Coherence that ties sentences of a text into a meaningfully connected structure is of great importance to text generation and translation. In this paper, we propose a topic-based coherence model to produce coherence for document translation, in terms of the continuity of sentence topics in a text. We automatically extract a coherence chain for each source text to be translated. Based on the extracted source coherence chain, we adopt a maximum entropy classifier to predict the target coherence chain that defines a linear topic structure for the target document. The proposed topic-based coherence model then uses the predicted target coherence chain to help decoder select coherent word/phrase translations. Our experiments show that incorporating the topic-based coherence model into machine translation achieves substantial improvement over both the baseline and previous methods that integrate document topics rather than coherence chains into machine translation.

IJCAI Conference 2013 Conference Paper

Modeling Lexical Cohesion for Document-Level Machine Translation

• Deyi Xiong
• Guosheng Ben
• Min Zhang
• Yajuan Lü
• Qun Liu

Lexical cohesion arises from a chain of lexical items that establish links between sentences in a text. In this paper we propose three different models to capture lexical cohesion for document-level machine translation: (a) a direct reward model where translation hypotheses are rewarded whenever lexical cohesion devices occur in them, (b) a conditional probability model where the appropriateness of using lexical cohesion devices is measured, and (c) a mutual information trigger model where a lexical cohesion relation is considered as a trigger pair and the strength of the association between the trigger and the triggered item is estimated by mutual information. We integrate the three models into hierarchical phrase-based machine translation and evaluate their effectiveness on the NIST Chinese-English translation tasks with large-scale training data. Experiment results show that all three models can achieve substantial improvements over the baseline and that the mutual information trigger model performs better than the others.

IJCAI Conference 2013 Conference Paper

Smoothing for Bracketing Induction

• Xiangyu Duan
• Min Zhang
• Wenliang Chen

Bracketing induction is the unsupervised learning of hierarchical constituents without labeling their syntactic categories such as verb phrase (VP) from natural raw sentences. Constituent Context Model (CCM) is an effective generative model for the bracketing induction, but the CCM computes probability of a constituent in a very straightforward way no matter how long this constituent is. Such method causes severe data sparse problem because long constituents are more unlikely to appear in test set. To overcome the data sparse problem, this paper proposes to define a non-parametric Bayesian prior distribution, namely the Pitman-Yor Process (PYP) prior, over constituents for constituent smoothing. The PYP prior functions as a back-off smoothing method through using a hierarchical smoothing scheme (HSS). Various kinds of HSS are proposed in this paper. We find that two kinds of HSS are effective, attaining or significantly improving the state-ofthe-art performance of the bracketing induction evaluated on standard treebanks of various languages, while another kind of HSS, which is commonly used for smoothing sequences by ngram Markovization, is not effective for improving the performance of the CCM.

YNIMG Journal 2012 Journal Article

A prior feature SVM-MRF based method for mouse brain segmentation

• Teresa Wu
• Min Hyeok Bae
• Min Zhang
• Rong Pan
• Alexandra Badea

We introduce an automated method, called prior feature Support Vector Machine-Markov Random Field (pSVMRF), to segment three-dimensional mouse brain Magnetic Resonance Microscopy (MRM) images. Our earlier work, extended MRF (eMRF) integrated Support Vector Machine (SVM) and Markov Random Field (MRF) approaches, leading to improved segmentation accuracy; however, the computation of eMRF is very expensive, which may limit its performance on segmentation and robustness. In this study pSVMRF reduces training and testing time for SVM, while boosting segmentation performance. Unlike the eMRF approach, where MR intensity information and location priors are linearly combined, pSVMRF combines this information in a nonlinear fashion, and enhances the discriminative ability of the algorithm. We validate the proposed method using MR imaging of unstained and actively stained mouse brain specimens, and compare segmentation accuracy with two existing methods: eMRF and MRF. C57BL/6 mice are used for training and testing, using cross validation. For formalin fixed C57BL/6 specimens, pSVMRF outperforms both eMRF and MRF. The segmentation accuracy for C57BL/6 brains, stained or not, was similar for larger structures like hippocampus and caudate putamen, (~87%), but increased substantially for smaller regions like susbtantia nigra (from 78. 36% to 91. 55%), and anterior commissure (from ~50% to ~80%). To test segmentation robustness against increased anatomical variability we add two strains, BXD29 and a transgenic mouse model of Alzheimer's disease. Segmentation accuracy for new strains is 80% for hippocampus, and caudate putamen, indicating that pSVMRF is a promising approach for phenotyping mouse models of human brain disorders.

IJCAI Conference 2011 Conference Paper

Predicting Epidemic Tendency through Search Behavior Analysis

• Danqing Xu
• Yiqun Liu
• Min Zhang
• Shaoping Ma
• Anqi Cui
• Liyun Ru

The possibility that influenza activity can be generally detected through search log analysis has been explored in recent years. However, previous studies have mainly focused on influenza, and little attention has been paid to other epidemics. With an analysis of web user behavior data, we consider the problem of predicting the tendency of hand-foot -and-mouth disease (HFMD), whose out-break in 2010 resulted in a great panic in China. In addi-tion to search queries, we consider users' interactions with search engines. Given the collected search logs, we cluster HFMD-related search queries, medical pages and news reports into the following sets: epidemic-related queries (ERQs), epidemic-related pages (ERPs) and ep-idemic-related news (ERNs). Furthermore, we count their own frequencies as different features, and we conduct a regression analysis with current HFMD occurrences. The experimental results show that these features exhibit good performances on both accuracy and timeliness.

AAAI Conference 2011 Conference Paper

Tree Sequence Kernel for Natural Language

• Jun Sun
• Min Zhang
• Chew Lim Tan

We propose Tree Sequence Kernel (TSK), which implicitly exhausts the structure features of a sequence of subtrees embedded in the phrasal parse tree. By incorporating the capability of sequence kernel, TSK enriches tree kernel with tree sequence features so that it may provide additional useful patterns for machine learning applications. Two approaches of penalizing the substructures are proposed and both can be accomplished by efficient algorithms via dynamic programming. Evaluations are performed on two natural language tasks, i. e. Question Classification and Relation Extraction. Experimental results suggest that TSK outperforms tree kernel for both tasks, which also reveals that the structure features made up of multiple subtrees are effective and play a complementary role to the single tree structure.

TCS Journal 2008 Journal Article

Computational self-assembly

• Pierre-Louis Curien
• Vincent Danos
• Jean Krivine
• Min Zhang

The object of this paper is to probe the computational limits of an applied concurrent language called κ. This language describes how agents can bind and modify each other. It is meant as a syntactic medium to build, discuss and execute descriptions of cellular signalling pathways. However, it can be studied independently of its intended interpretation, and this is what we are doing here. Specifically, we define a reduction of κ to a fragment where interactions can involve at most two agents at a time. The translation relies on an implicit causality analysis which permits escaping deadlocks. It incurs only a linear blow up in the number of rules. Its correctness is spelt out in terms of the existence of a specific weak bisimulation and is proved in detail. To compensate for the binary restriction, one allows components to create unique names. When using acyclic rules, this additional facility of name creation is not needed and κ can be reduced to a binary form as is.