Wu-Jun Li

AAAI Conference 2026 Conference Paper

Controllable Financial Market Generation with Diffusion Guided Meta Agent

• Yu-Hao Huang
• Chang Xu
• Yang Liu
• Weiqing Liu
• Wu-Jun Li
• Jiang Bian

Generative modeling has transformed many fields, such as language and visual modeling, while its application in financial markets remains under-explored. As the minimal unit within a financial market is an order, order-flow modeling represents a fundamental generative financial task. However, current approaches often yield unsatisfactory fidelity in generating order flow, and their generation lacks controllability, thereby limiting their practical applications. In this paper, we formulate the challenge of controllable financial market generation, and propose a Diffusion Guided Meta Agent (DigMA) model to address it. Specifically, we employ a conditional diffusion model to capture the dynamics of the market state represented by time-evolving distribution parameters of the mid-price return rate and the order arrival rate, and we define a meta agent with financial economic priors to generate orders from the corresponding distributions. Extensive experimental results show that DigMA achieves superior controllability and generation fidelity. Moreover, we validate its effectiveness as a generative environment for downstream high-frequency trading tasks and its computational efficiency.

AAAI Conference 2026 Conference Paper

Ordered Local Momentum for Asynchronous Distributed Learning Under Arbitrary Delays

• Chang-Wei Shi
• Shi-Shang Wang
• Wu-Jun Li

Momentum SGD (MSGD) serves as a foundational optimizer in training deep models due to momentum's key role in accelerating convergence and enhancing generalization. Meanwhile, asynchronous distributed learning is crucial for training large-scale deep models, especially when the computing capabilities of the workers in the cluster are heterogeneous. To reduce communication frequency, local updates are widely adopted in distributed learning. However, how to implement asynchronous distributed MSGD with local updates remains unexplored. To solve this problem, we propose a novel method, called ordered local momentum (OrLoMo), for asynchronous distributed learning. In OrLoMo, each worker runs MSGD locally. Then the local momentum from each worker will be aggregated by the server in order based on its global iteration index. To the best of our knowledge, OrLoMo is the first method to implement asynchronous distributed MSGD with local updates. We prove the convergence of OrLoMo for non-convex problems under arbitrary delays. Experiments validate that OrLoMo can outperform its synchronous counterpart and other asynchronous methods.

AAAI Conference 2025 Conference Paper

DrugHash: Hashing Based Contrastive Learning for Virtual Screening

• Jin Han
• Yun Hong
• Wu-Jun Li

Virtual screening (VS) is a critical step in computer-aided drug discovery, aiming to identify molecules that bind to a specific target protein. Traditional VS methods, such as docking, are often too time-consuming to efficiently screen large-scale molecular databases. Recent advances in deep learning have demonstrated that learning vector representations for both proteins and molecules using contrastive learning can outperform traditional docking methods. However, considering that the target databases often contain billions of molecules, real-valued vector representations adopted by existing methods can still incur large memory and time cost in VS. To address this problem, we propose DrugHash, a hashing-based contrastive learning method for VS. DrugHash formulates VS as a retrieval task that leverages binary hash codes for efficient retrieval. In particular, DrugHash designs a simple yet effective hashing strategy to enable end-to-end learning of binary hash codes for both proteins and molecules, which can dramatically reduce the memory and time cost with higher accuracy compared with existing methods. Experimental results show that DrugHash can outperform existing methods to achieve state-of-the-art accuracy, with at least a 32 times reduction in memory cost and a 4.6 times improvement in speed.

NeurIPS Conference 2025 Conference Paper

Gated Integration of Low-Rank Adaptation for Continual Learning of Large Language Models

• Yan-Shuo Liang
• Jia-Rui Chen
• Wu-Jun Li

Continual learning (CL), which requires the model to learn multiple tasks sequentially, is crucial for large language models (LLMs). Recently, low-rank adaptation (LoRA), one of the most representative parameter-efficient fine-tuning (PEFT) methods, has gained increasing attention in CL of LLMs. However, most existing CL methods based on LoRA typically expand a new LoRA branch to learn each new task and force the new and old LoRA branches to influence old tasks equally, potentially leading to forgetting. In this work, we propose a new method, called gated integration of low-rank adaptation (GainLoRA), for CL of LLMs. GainLoRA expands a new LoRA branch for each new task and introduces gating modules to integrate the new and old LoRA branches. Furthermore, GainLoRA leverages the new gating module to minimize the influence from the new LoRA branch to old tasks, effectively mitigating forgetting and improving the model's overall performance. Experimental results on CL benchmarks demonstrate that GainLoRA outperforms existing state-of-the-art methods.

ICML Conference 2025 Conference Paper

On the Tension between Byzantine Robustness and No-Attack Accuracy in Distributed Learning

• Yi-Rui Yang
• Chang-Wei Shi
• Wu-Jun Li

Byzantine-robust distributed learning (BRDL), which refers to distributed learning that can work with potential faulty or malicious workers (also known as Byzantine workers), has recently attracted much research attention. Robust aggregators are widely used in existing BRDL methods to obtain robustness against Byzantine workers. However, Byzantine workers do not always exist in applications. As far as we know, there is almost no existing work theoretically investigating the effect of using robust aggregators when there are no Byzantine workers. To bridge this knowledge gap, we theoretically analyze the aggregation error for robust aggregators when there are no Byzantine workers. Specifically, we show that the worst-case aggregation error without Byzantine workers increases with the increase of the number of Byzantine workers that a robust aggregator can tolerate. The theoretical result reveals the tension between Byzantine robustness and no-attack accuracy, which refers to accuracy without faulty workers and malicious workers in this paper. Furthermore, we provide lower bounds for the convergence rate of gradient descent with robust aggregators for non-convex objective functions and objective functions that satisfy the Polyak-Lojasiewicz (PL) condition, respectively. We also prove the tightness of the lower bounds. The lower bounds for convergence rate reveal similar tension between Byzantine robustness and no-attack accuracy. Empirical results further support our theoretical findings.

AAAI Conference 2025 Conference Paper

TimeDP: Learning to Generate Multi-Domain Time Series with Domain Prompts

• Yu-Hao Huang
• Chang Xu
• Yueying Wu
• Wu-Jun Li
• Jiang Bian

Time series generation models are crucial for applications like data augmentation and privacy preservation. Most existing time series generation models are typically designed to generate data from one specified domain. While leveraging data from other domain for better generalization is proved to work in other application areas, this approach remains challenging for time series modeling due to the large divergence in patterns among different real world time series categories. In this paper, we propose a multi-domain time series diffusion model with domain prompts, named TimeDP. In TimeDP, we utilize a time series semantic prototype module which defines time series prototypes to represent time series basis, each prototype vector serving as "word" representing some elementary time series feature. A prototype assignment module is applied to extract the extract domain specific prototype weights, for learning domain prompts as generation condition. During sampling, we extract ``domain prompt" with few-shot samples from the target domain and use the domain prompts as condition to generate time series samples. Experiments demonstrate that our method outperforms baselines to provide the state-of-the-art in-domain generation quality and strong unseen domain generation capability.

ICLR Conference 2024 Conference Paper

On the Effect of Batch Size in Byzantine-Robust Distributed Learning

• Yi-Rui Yang
• Chang-Wei Shi
• Wu-Jun Li

Byzantine-robust distributed learning (BRDL), in which computing devices are likely to behave abnormally due to accidental failures or malicious attacks, has recently become a hot research topic. However, even in the independent and identically distributed (i.i.d.) case, existing BRDL methods will suffer a significant drop on model accuracy due to the large variance of stochastic gradients. Increasing batch sizes is a simple yet effective way to reduce the variance. However, when the total number of gradient computation is fixed, a too-large batch size will lead to a too-small iteration number (update number), which may also degrade the model accuracy. In view of this challenge, we mainly study the effect of batch size when the total number of gradient computation is fixed in this work. In particular, we show that when the total number of gradient computation is fixed, the optimal batch size corresponding to the tightest theoretical upper bound in BRDL increases with the fraction of Byzantine workers. Therefore, compared to the case without attacks, a larger batch size is preferred when under Byzantine attacks. Motivated by the theoretical finding, we propose a novel method called Byzantine-robust stochastic gradient descent with normalized momentum (ByzSGDnm) in order to further increase model accuracy in BRDL. We theoretically prove the convergence of ByzSGDnm for general non-convex cases under Byzantine attacks. Empirical results show that when under Byzantine attacks, compared to the cases of small batch sizes, setting a relatively large batch size can significantly increase the model accuracy, which is consistent with our theoretical results. Moreover, ByzSGDnm can achieve higher model accuracy than existing BRDL methods when under deliberately crafted attacks. In addition, we empirically show that increasing batch sizes has the bonus of training acceleration.

NeurIPS Conference 2024 Conference Paper

Ordered Momentum for Asynchronous SGD

• Chang-Wei Shi
• Yi-Rui Yang
• Wu-Jun Li

Distributed learning is essential for training large-scale deep models. Asynchronous SGD (ASGD) and its variants are commonly used distributed learning methods, particularly in scenarios where the computing capabilities of workers in the cluster are heterogeneous. Momentum has been acknowledged for its benefits in both optimization and generalization in deep model training. However, existing works have found that naively incorporating momentum into ASGD can impede the convergence. In this paper, we propose a novel method called ordered momentum (OrMo) for ASGD. In OrMo, momentum is incorporated into ASGD by organizing the gradients in order based on their iteration indexes. We theoretically prove the convergence of OrMo with both constant and delay-adaptive learning rates for non-convex problems. To the best of our knowledge, this is the first work to establish the convergence analysis of ASGD with momentum without dependence on the maximum delay. Empirical results demonstrate that OrMo can achieve better convergence performance compared with ASGD and other asynchronous methods with momentum.

ICRA Conference 2024 Conference Paper

Weighting Online Decision Transformer with Episodic Memory for Offline-to-Online Reinforcement Learning

• Xiao Ma
• Wu-Jun Li

Offline reinforcement learning (RL) has been shown to be successfully modeled as a sequence modeling problem, drawing inspiration from the success of Transformers. Offline RL is often limited by the quality of the offline dataset, so offline-to-online RL is a more realistic setting. Online decision transformer (ODT) is an effective and representative sequence modeling-based offline-to-online RL method. Despite its effectiveness, ODT still suffers from the sample inefficiency problem during the online fine-tuning phase. This sample inefficiency problem arises because the agent treats all state-action pairs in the replay buffer equally when trying to learn from the replay buffer. In this paper, we propose a simple yet effective method, called weighting online decision transformer with episodic memory (WODTEM), to improve sample efficiency. We first attempt to introduce an episodic memory (EM) mechanism into the sequence modeling-based RL methods. By utilizing the EM mechanism, we propose a novel training objective with a weighting function, based on ODT, to improve sample efficiency. Experimental results on multiple tasks show that WODTEM can improve sample efficiency.

JMLR Journal 2023 Journal Article

Buffered Asynchronous SGD for Byzantine Learning

• Yi-Rui Yang
• Wu-Jun Li

Distributed learning has become a hot research topic due to its wide application in cluster-based large-scale learning, federated learning, edge computing, and so on. Most traditional distributed learning methods typically assume no failure or attack. However, many unexpected cases, such as communication failure and even malicious attack, may happen in real applications. Hence, Byzantine learning (BL), which refers to distributed learning with failure or attack, has recently attracted much attention. Most existing BL methods are synchronous, which are impractical in some applications due to heterogeneous or offline workers. In these cases, asynchronous BL (ABL) is usually preferred. In this paper, we propose a novel method, called buffered asynchronous stochastic gradient descent (BASGD), for ABL. To the best of our knowledge, BASGD is the first ABL method that can resist non-omniscient attacks without storing any instances on the server. Furthermore, we also propose an improved variant of BASGD, called BASGD with momentum (BASGDm), by introducing local momentum into BASGD. Compared with those methods which need to store instances on server, BASGD and BASGDm have a wider scope of application. Both BASGD and BASGDm are compatible with various aggregation rules. Moreover, both BASGD and BASGDm are proved to be convergent and able to resist failure or attack. Empirical results show that our methods significantly outperform existing ABL baselines when there exists failure or attack on workers. [abs] [ pdf ][ bib ] &copy JMLR 2023. ( edit, beta )

AAMAS Conference 2023 Conference Paper

Grey-box Adversarial Attack on Communication in Multi-agent Reinforcement Learning

• Xiao Ma
• Wu-Jun Li

Although research on communication in multi-agent reinforcement learning (MARL) has achieved some progress, the vulnerability of the communication mechanism in MARL caused by adversarial communication messages generated by malicious agents has not been well investigated. Existing works about adversarial communication messages in MARL focus on the black-box scenario where the attacker cannot access any model information about the multiagent system (MAS). But a more practical setting is the grey-box scenario where the attacker can access the model information about its controlled agent. To the best of our knowledge, there has not been any work investigating grey-box attacks on communication in MARL. In this paper, we propose the first grey-box attack method on communication in MARL, which is called victim-simulation based adversarial attack (VSA). At each timestep, the attacker simulates a victim attacked by other regular agents’ communication messages and generates adversarial perturbations on its received communication messages. The aggregation of these perturbations is sent by the attacker to the regular agents through communication messages, which will induce non-optimal actions of the regular agents. Experimental results show that VSA can effectively degrade the performance of the MAS on Predator-Prey. The findings in this paper will make researchers aware of the grey-box attack in MARL.

NeurIPS Conference 2023 Conference Paper

Loss Decoupling for Task-Agnostic Continual Learning

• Yan-Shuo Liang
• Wu-Jun Li

Continual learning requires the model to learn multiple tasks in a sequential order. To perform continual learning, the model must possess the abilities to maintain performance on old tasks (stability) and adapt itself to learn new tasks (plasticity). Task-agnostic problem in continual learning is a challenging problem, in which task identities are not available in the inference stage and hence the model must learn to distinguish all the classes in all the tasks. In task-agnostic problem, the model needs to learn two new objectives for learning a new task, including distinguishing new classes from old classes and distinguishing between different new classes. For task-agnostic problem, replay-based methods are commonly used. These methods update the model with both saved old samples and new samples for continual learning. Most existing replay-based methods mix the two objectives in task-agnostic problem together, inhibiting the models from achieving a good trade-off between stability and plasticity. In this paper, we propose a simple yet effective method, called loss decoupling (LODE), for task-agnostic continual learning. LODE separates the two objectives for the new task by decoupling the loss of the new task. As a result, LODE can assign different weights for different objectives, which provides a way to obtain a better trade-off between stability and plasticity than those methods with coupled loss. Experiments show that LODE can outperform existing state-of-the-art replay-based methods on multiple continual learning datasets.

ICRA Conference 2022 Conference Paper

Cross Domain Robot Imitation with Invariant Representation

• Zhao-Heng Yin
• Lingfeng Sun
• Hengbo Ma
• Masayoshi Tomizuka
• Wu-Jun Li

Animals are able to imitate each others' behavior, despite their difference in biomechanics. In contrast, imitating other similar robots is a much more challenging task in robotics. This problem is called cross domain imitation learning (CDIL). In this paper, we consider CDIL on a class of similar robots. We tackle this problem by introducing an imitation learning algorithm based on invariant representation. We propose to learn invariant state and action representations, which align the behavior of multiple robots so that CDIL becomes possible. Compared with previous invariant representation learning methods for similar purposes, our method does not require human-labeled pairwise data for training. Instead, we use cycle-consistency and domain confusion to align the representation and increase its robustness. We test the algorithm on multiple robots in the simulator and show that unseen new robot instances can be trained with existing expert demonstrations successfully. Qualitative results also demonstrate that the proposed method is able to learn similar representations for different robots with similar behaviors, which is essential for successful CDIL.

ICML Conference 2021 Conference Paper

BASGD: Buffered Asynchronous SGD for Byzantine Learning

• Yi-Rui Yang
• Wu-Jun Li

Distributed learning has become a hot research topic due to its wide application in cluster-based large-scale learning, federated learning, edge computing and so on. Most traditional distributed learning methods typically assume no failure or attack. However, many unexpected cases, such as communication failure and even malicious attack, may happen in real applications. Hence, Byzantine learning (BL), which refers to distributed learning with failure or attack, has recently attracted much attention. Most existing BL methods are synchronous, which are impractical in some applications due to heterogeneous or offline workers. In these cases, asynchronous BL (ABL) is usually preferred. In this paper, we propose a novel method, called buffered asynchronous stochastic gradient descent (BASGD), for ABL. To the best of our knowledge, BASGD is the first ABL method that can resist malicious attack without storing any instances on server. Compared with those methods which need to store instances on server, BASGD has a wider scope of application. BASGD is proved to be convergent, and be able to resist failure or attack. Empirical results show that BASGD significantly outperforms vanilla asynchronous stochastic gradient descent (ASGD) and other ABL baselines when there exists failure or attack on workers.

IJCAI Conference 2021 Conference Paper

Blocking-based Neighbor Sampling for Large-scale Graph Neural Networks

• Kai-Lang Yao
• Wu-Jun Li

The exponential increase in computation and memory complexity with the depth of network has become the main impediment to the successful application of graph neural networks (GNNs) on large-scale graphs like graphs with hundreds of millions of nodes. In this paper, we propose a novel neighbor sampling strategy, dubbed blocking-based neighbor sampling (BNS), for efficient training of GNNs on large-scale graphs. Specifically, BNS adopts a policy to stochastically block the ongoing expansion of neighboring nodes, which can reduce the rate of the exponential increase in computation and memory complexity of GNNs. Furthermore, a reweighted policy is applied to graph convolution, to adjust the contribution of blocked and non-blocked neighbors to central nodes. We theoretically prove that BNS provides an unbiased estimation for the original graph convolution operation. Extensive experiments on three benchmark datasets show that, on large-scale graphs, BNS is 2X~5X faster than state-of-the-art methods when achieving the same accuracy. Moreover, even on the small-scale graphs, BNS also demonstrates the advantage of low time cost.

AAAI Conference 2020 Conference Paper

Hashing Based Answer Selection

• Dong Xu
• Wu-Jun Li

Answer selection is an important subtask of question answering (QA), in which deep models usually achieve better performance than non-deep models. Most deep models adopt question-answer interaction mechanisms, such as attention, to get vector representations for answers. When these interaction based deep models are deployed for online prediction, the representations of all answers need to be recalculated for each question. This procedure is time-consuming for deep models with complex encoders like BERT which usually have better accuracy than simple encoders. One possible solution is to store the matrix representation (encoder output) of each answer in memory to avoid recalculation. But this will bring large memory cost. In this paper, we propose a novel method, called hashing based answer selection (HAS), to tackle this problem. HAS adopts a hashing strategy to learn a binary matrix representation for each answer, which can dramatically reduce the memory cost for storing the matrix representations of answers. Hence, HAS can adopt complex encoders like BERT in the model, but the online prediction of HAS is still fast with a low memory cost. Experimental results on three popular answer selection datasets show that HAS can outperform existing models to achieve state-of-the-art performance.

JMLR Journal 2019 Journal Article

Robust Frequent Directions with Application in Online Learning

• Luo Luo
• Cheng Chen
• Zhihua Zhang
• Wu-Jun Li
• Tong Zhang

The frequent directions (FD) technique is a deterministic approach for online sketching that has many applications in machine learning. The conventional FD is a heuristic procedure that often outputs rank deficient matrices. To overcome the rank deficiency problem, we propose a new sketching strategy called robust frequent directions (RFD) by introducing a regularization term. RFD can be derived from an optimization problem. It updates the sketch matrix and the regularization term adaptively and jointly. RFD reduces the approximation error of FD without increasing the computational cost. We also apply RFD to online learning and propose an effective hyperparameter-free online Newton algorithm. We derive a regret bound for our online Newton algorithm based on RFD, which guarantees the robustness of the algorithm. The experimental studies demonstrate that the proposed method outperforms state-of-the-art second order online learning algorithms. [abs] [ pdf ][ bib ] &copy JMLR 2019. ( edit, beta )

AAAI Conference 2018 Conference Paper

Asymmetric Deep Supervised Hashing

• Qing-Yuan Jiang
• Wu-Jun Li

Hashing has been widely used for large-scale approximate nearest neighbor search because of its storage and search ef- ficiency. Recent work has found that deep supervised hashing can significantly outperform non-deep supervised hashing in many applications. However, most existing deep supervised hashing methods adopt a symmetric strategy to learn one deep hash function for both query points and database (retrieval) points. The training of these symmetric deep supervised hashing methods is typically time-consuming, which makes them hard to effectively utilize the supervised information for cases with large-scale database. In this paper, we propose a novel deep supervised hashing method, called asymmetric deep supervised hashing (ADSH), for large-scale nearest neighbor search. ADSH treats the query points and database points in an asymmetric way. More specifically, ADSH learns a deep hash function only for query points, while the hash codes for database points are directly learned. The training of ADSH is much more efficient than that of traditional symmetric deep supervised hashing methods. Experiments show that ADSH can achieve state-of-the-art performance in real applications.

NeurIPS Conference 2018 Conference Paper

Proximal SCOPE for Distributed Sparse Learning

• Shenyi Zhao
• Gong-Duo Zhang
• Ming-Wei Li
• Wu-Jun Li

Distributed sparse learning with a cluster of multiple machines has attracted much attention in machine learning, especially for large-scale applications with high-dimensional data. One popular way to implement sparse learning is to use L1 regularization. In this paper, we propose a novel method, called proximal SCOPE (pSCOPE), for distributed sparse learning with L1 regularization. pSCOPE is based on a cooperative autonomous local learning (CALL) framework. In the CALL framework of pSCOPE, we find that the data partition affects the convergence of the learning procedure, and subsequently we define a metric to measure the goodness of a data partition. Based on the defined metric, we theoretically prove that pSCOPE is convergent with a linear convergence rate if the data partition is good enough. We also prove that better data partition implies faster convergence rate. Furthermore, pSCOPE is also communication efficient. Experimental results on real data sets show that pSCOPE can outperform other state-of-the-art distributed methods for sparse learning.

AAAI Conference 2017 Conference Paper

Lock-Free Optimization for Non-Convex Problems

• Shen-Yi Zhao
• Gong-Duo Zhang
• Wu-Jun Li

Stochastic gradient descent (SGD) and its variants have attracted much attention in machine learning due to their ef- ficiency and effectiveness for optimization. To handle largescale problems, researchers have recently proposed several lock-free strategy based parallel SGD (LF-PSGD) methods for multi-core systems. However, existing works have only proved the convergence of these LF-PSGD methods for convex problems. To the best of our knowledge, no work has proved the convergence of the LF-PSGD methods for nonconvex problems. In this paper, we provide the theoretical proof about the convergence of two representative LF-PSGD methods, Hogwild! and AsySVRG, for non-convex problems. Empirical results also show that both Hogwild! and AsySVRG are convergent on non-convex problems, which successfully verifies our theoretical results.

AAAI Conference 2017 Conference Paper

SCOPE: Scalable Composite Optimization for Learning on Spark

• Shen-Yi Zhao
• Ru Xiang
• Ying-Hao Shi
• Peng Gao
• Wu-Jun Li

Many machine learning models, such as logistic regression (LR) and support vector machine (SVM), can be formulated as composite optimization problems. Recently, many distributed stochastic optimization (DSO) methods have been proposed to solve the large-scale composite optimization problems, which have shown better performance than traditional batch methods. However, most of these DSO methods might not be scalable enough. In this paper, we propose a novel DSO method, called scalable composite optimization for learning (SCOPE), and implement it on the fault-tolerant distributed platform Spark. SCOPE is both computation-efficient and communication-efficient. Theoretical analysis shows that SCOPE is convergent with linear convergence rate when the objective function is strongly convex. Furthermore, empirical results on real datasets show that SCOPE can outperform other state-of-the-art distributed learning methods on Spark, including both batch learning methods and DSO methods.

IJCAI Conference 2017 Conference Paper

Semi-Supervised Deep Hashing with a Bipartite Graph

• Xinyu Yan
• Lijun Zhang
• Wu-Jun Li

Recently, deep learning has been successfully applied to the problem of hashing, yielding remarkable performance compared to traditional methods with hand-crafted features. However, most of existing deep hashing methods are designed for the supervised scenario and require a large number of labeled data. In this paper, we propose a novel semi-supervised hashing method for image retrieval, named Deep Hashing with a Bipartite Graph (DHBG), to simultaneously learn embeddings, features and hash codes. More specifically, we construct a bipartite graph to discover the underlying structure of data, based on which an embedding is generated for each instance. Then, we feed raw pixels as well as embeddings to a deep neural network, and concatenate the resulting features to determine the hash code. Compared to existing methods, DHBG is a universal framework that is able to utilize various types of graphs and losses. Furthermore, we propose an inductive variant of DHBG to support out-of-sample extensions. Experimental results on real datasets show that our DHBG outperforms state-of-the-art hashing methods.

AAAI Conference 2016 Conference Paper

Column Sampling Based Discrete Supervised Hashing

• Wang-Cheng Kang
• Wu-Jun Li
• Zhi-Hua Zhou

By leveraging semantic (label) information, supervised hashing has demonstrated better accuracy than unsupervised hashing in many real applications. Because the hashing-code learning problem is essentially a discrete optimization problem which is hard to solve, most existing supervised hashing methods try to solve a relaxed continuous optimization problem by dropping the discrete constraints. However, these methods typically suffer from poor performance due to the errors caused by the relaxation. Some other methods try to directly solve the discrete optimization problem. However, they are typically time-consuming and unscalable. In this paper, we propose a novel method, called column sampling based discrete supervised hashing (COSDISH), to directly learn the discrete hashing code from semantic information. COSDISH is an iterative method, in each iteration of which several columns are sampled from the semantic similarity matrix and then the hashing code is decomposed into two parts which can be alternately optimized in a discrete way. Theoretical analysis shows that the learning (optimization) algorithm of COSDISH has a constant-approximation bound in each step of the alternating optimization procedure. Empirical results on datasets with semantic labels illustrate that COSDISH can outperform the state-of-the-art methods in real applications like image retrieval.

AAAI Conference 2016 Conference Paper

Fast Asynchronous Parallel Stochastic Gradient Descent: A Lock-Free Approach with Convergence Guarantee

• Shen-Yi Zhao
• Wu-Jun Li

Stochastic gradient descent (SGD) and its variants have become more and more popular in machine learning due to their efficiency and effectiveness. To handle large-scale problems, researchers have recently proposed several parallel SGD methods for multicore systems. However, existing parallel SGD methods cannot achieve satisfactory performance in real applications. In this paper, we propose a fast asynchronous parallel SGD method, called AsySVRG, by designing an asynchronous strategy to parallelize the recently proposed SGD variant called stochastic variance reduced gradient (SVRG). AsySVRG adopts a lock-free strategy which is more efficient than other strategies with locks. Furthermore, we theoretically prove that AsySVRG is convergent with a linear convergence rate. Both theoretical and empirical results show that AsySVRG can outperform existing state-ofthe-art parallel SGD methods like Hogwild! in terms of convergence rate and computation cost.

IJCAI Conference 2016 Conference Paper

Feature Learning Based Deep Supervised Hashing with Pairwise Labels

• Wu-Jun Li
• Sheng Wang
• Wang-Cheng Kang

Recent years have witnessed wide application of hashing for large-scale image retrieval. However, most existing hashing methods are based on hand-crafted features which might not be optimally compatible with the hashing procedure. Recently, deep hashing methods have been proposed to perform simultaneous feature learning and hash-code learning with deep neural networks, which have shown better performance than traditional hashing methods with hand-crafted features. Most of these deep hashing methods are supervised whose supervised information is given with triplet labels. For another common application scenario with pairwise labels, there have not existed methods for simultaneous feature learning and hash-code learning. In this paper, we propose a novel deep hashing method, called deep pairwise-supervised hashing (DPSH), to perform simultaneous feature learning and hash-code learning for applications with pairwise labels. Experiments on real datasets show that our DPSH method can outperform other methods to achieve the state-of-the-art performance in image retrieval applications.

IJCAI Conference 2015 Conference Paper

Scalable Graph Hashing with Feature Transformation

• Qing-Yuan Jiang
• Wu-Jun Li

Hashing has been widely used for approximate nearest neighbor (ANN) search in big data applications because of its low storage cost and fast retrieval speed. The goal of hashing is to map the data points from the original space into a binary-code space where the similarity (neighborhood structure) in the original space is preserved. By directly exploiting the similarity to guide the hashing code learning procedure, graph hashing has attracted much attention. However, most existing graph hashing methods cannot achieve satisfactory performance in real applications due to the high complexity for graph modeling. In this paper, we propose a novel method, called scalable graph hashing with feature transformation (SGH), for large-scale graph hashing. Through feature transformation, we can effectively approximate the whole graph without explicitly computing the similarity graph matrix, based on which a sequential learning method is proposed to learn the hash functions in a bit-wise manner. Experiments on two datasets with one million data points show that our SGH method can outperform the state-of-the-art methods in terms of both accuracy and scalability.

ICML Conference 2015 Conference Paper

Support Matrix Machines

• Luo Luo
• Yubo Xie
• Zhihua Zhang 0004
• Wu-Jun Li

In many classification problems such as electroencephalogram (EEG) classification and image classification, the input features are naturally represented as matrices rather than vectors or scalars. In general, the structure information of the original feature matrix is useful and informative for data analysis tasks such as classification. One typical structure information is the correlation between columns or rows in the feature matrix. To leverage this kind of structure information, we propose a new classification method that we call support matrix machine (SMM). Specifically, SMM is defined as a hinge loss plus a so-called spectral elastic net penalty which is a spectral extension of the conventional elastic net over a matrix. The spectral elastic net enjoys a property of grouping effect, i. e. , strongly correlated columns or rows tend to be selected altogether or not. Since the optimization problem for SMM is convex, this encourages us to devise an alternating direction method of multipliers algorithm for solving the problem. Experimental results on EEG and face image classification data show that our model is more robust and efficient than the state-of-the-art methods.

ICML Conference 2014 Conference Paper

Coupled Group Lasso for Web-Scale CTR Prediction in Display Advertising

• Ling Yan
• Wu-Jun Li
• Gui-Rong Xue
• Dingyi Han

In display advertising, click through rate(CTR) prediction is the problem of estimating the probability that an advertisement (ad) is clicked when displayed to a user in a specific context. Due to its easy implementation and promising performance, logistic regression(LR) model has been widely used for CTR prediction, especially in industrial systems. However, it is not easy for LR to capture the nonlinear information, such as the conjunction information, from user features and ad features. In this paper, we propose a novel model, called coupled group lasso(CGL), for CTR prediction in display advertising. CGL can seamlessly integrate the conjunction information from user features and ad features for modeling. Furthermore, CGL can automatically eliminate useless features for both users and ads, which may facilitate fast online prediction. Scalability of CGL is ensured through feature hashing and distributed implementation. Experimental results on real-world data sets show that our CGL model can achieve state-of-the-art performance on web-scale CTR prediction tasks.

NeurIPS Conference 2014 Conference Paper

Distributed Power-law Graph Computing: Theoretical and Empirical Analysis

• Cong Xie
• Ling Yan
• Wu-Jun Li
• Zhihua Zhang

With the emergence of big graphs in a variety of real applications like social networks, machine learning based on distributed graph-computing~(DGC) frameworks has attracted much attention from big data machine learning community. In DGC frameworks, the graph partitioning~(GP) strategy plays a key role to affect the performance, including the workload balance and communication cost. Typically, the degree distributions of natural graphs from real applications follow skewed power laws, which makes GP a challenging task. Recently, many methods have been proposed to solve the GP problem. However, the existing GP methods cannot achieve satisfactory performance for applications with power-law graphs. In this paper, we propose a novel vertex-cut method, called \emph{degree-based hashing}~(DBH), for GP. DBH makes effective use of the skewed degree distributions for GP. We theoretically prove that DBH can achieve lower communication cost than existing methods and can simultaneously guarantee good workload balance. Furthermore, empirical results on several large power-law graphs also show that DBH can outperform the state of the art.

AAAI Conference 2014 Conference Paper

Large-Scale Supervised Multimodal Hashing with Semantic Correlation Maximization

• Dongqing Zhang
• Wu-Jun Li

Due to its low storage cost and fast query speed, hashing has been widely adopted for similarity search in multimedia data. In particular, more and more attentions have been payed to multimodal hashing for search in multimedia data with multiple modalities, such as images with tags. Typically, supervised information of semantic labels is also available for the data points in many real applications. Hence, many supervised multimodal hashing (SMH) methods have been proposed to utilize such semantic labels to further improve the search accuracy. However, the training time complexity of most existing SMH methods is too high, which makes them unscalable to large-scale datasets. In this paper, a novel SMH method, called semantic correlation maximization (SCM), is proposed to seamlessly integrate semantic labels into the hashing learning procedure for large-scale data modeling. Experimental results on two real-world datasets show that SCM can significantly outperform the state-of-the-art SMH methods, in terms of both accuracy and scalability.

AIJ Journal 2014 Journal Article

Multicategory large margin classification methods: Hinge losses vs. coherence functions

• Zhihua Zhang
• Cheng Chen
• Guang Dai
• Wu-Jun Li
• Dit-Yan Yeung

Generalization of large margin classification methods from the binary classification setting to the more general multicategory setting is often found to be non-trivial. In this paper, we study large margin classification methods that can be seamlessly applied to both settings, with the binary setting simply as a special case. In particular, we explore the Fisher consistency properties of multicategory majorization losses and present a construction framework of majorization losses of the 0–1 loss. Under this framework, we conduct an in-depth analysis about three widely used multicategory hinge losses. Corresponding to the three hinge losses, we propose three multicategory majorization losses based on a coherence function. The limits of the three coherence losses as the temperature approaches zero are the corresponding hinge losses, and the limits of the minimizers of their expected errors are the minimizers of the expected errors of the corresponding hinge losses. Finally, we develop multicategory large margin classification methods by using a so-called multiclass C -loss.

IJCAI Conference 2013 Conference Paper

Collaborative Topic Regression with Social Regularization for Tag Recommendation

• Hao Wang
• Binyi Chen
• Wu-Jun Li

Recently, tag recommendation (TR) has become a very hot research topic in data mining and related areas. However, neither co-occurrence based methods which only use the item-tag matrix nor content based methods which only use the item content information can achieve satisfactory performance in real TR applications. Hence, how to effectively combine the item-tag matrix, item content information, and other auxiliary information into the same recommendation framework is the key challenge for TR. In this paper, we first adapt the collaborative topic regression (CTR) model, which has been successfully applied for article recommendation, to combine both item-tag matrix and item content information for TR. Furthermore, by extending CTR we propose a novel hierarchical Bayesian model, called CTR with social regularization (CTR-SR), to seamlessly integrate the item-tag matrix, item content information, and social networks between items into the same principled model. Experiments on real data demonstrate the effectiveness of our proposed models.

IJCAI Conference 2013 Conference Paper

Online Egocentric Models for Citation Networks

• Hao Wang
• Wu-Jun Li

With the emergence of large-scale evolving (timevarying) networks, dynamic network analysis (DNA) has become a very hot research topic in recent years. Although a lot of DNA methods have been proposed by researchers from different communities, most of them can only model snapshot data recorded at a very rough temporal granularity. Recently, some models have been proposed for DNA which can be used to model large-scale citation networks at a fine temporal granularity. However, they suffer from a significant decrease of accuracy over time because the learned parameters or node features are static (fixed) during the prediction process for evolving citation networks. In this paper, we propose a novel model, called online egocentric model (OEM), to learn time-varying parameters and node features for evolving citation networks. Experimental results on real-world citation networks show that our OEM can not only prevent the prediction accuracy from decreasing over time but also uncover the evolution of topics in citation networks.

AAAI Conference 2012 Conference Paper

Double-Bit Quantization for Hashing

• Weihao Kong
• Wu-Jun Li

Hashing, which tries to learn similarity-preserving binary codes for data representation, has been widely used for efficient nearest neighbor search in massive databases due to its fast query speed and low storage cost. Because it is NP hard to directly compute the best binary codes for a given data set, mainstream hashing methods typically adopt a two-stage strategy. In the first stage, several projected dimensions of real values are generated. Then in the second stage, the real values will be quantized into binary codes by thresholding. Currently, most existing methods use one single bit to quantize each projected dimension. One problem with this single-bit quantization (SBQ) is that the threshold typically lies in the region of the highest point density and consequently a lot of neighboring points close to the threshold will be hashed to totally different bits, which is unexpected according to the principle of hashing. In this paper, we propose a novel quantization strategy, called double-bit quantization (DBQ), to solve the problem of SBQ. The basic idea of DBQ is to quantize each projected dimension into double bits with adaptively learned thresholds. Extensive experiments on two real data sets show that our DBQ strategy can significantly outperform traditional SBQ strategy for hashing.

AAAI Conference 2012 Conference Paper

Emoticon Smoothed Language Models for Twitter Sentiment Analysis

• Kun-Lin Liu
• Wu-Jun Li
• Minyi Guo

Twitter sentiment analysis (TSA) has become a hot research topic in recent years. The goal of this task is to discover the attitude or opinion of the tweets, which is typically formulated as a machine learning based text classification problem. Some methods use manually labeled data to train fully supervised models, while others use some noisy labels, such as emoticons and hashtags, for model training. In general, we can only get a limited number of training data for the fully supervised models because it is very labor-intensive and time-consuming to manually label the tweets. As for the models with noisy labels, it is hard for them to achieve satisfactory performance due to the noise in the labels although it is easy to get a large amount of data for training. Hence, the best strategy is to utilize both manually labeled data and noisy labeled data for training. However, how to seamlessly integrate these two different kinds of data into the same learning framework is still a challenge. In this paper, we present a novel model, called emoticon smoothed language model (ESLAM), to handle this challenge. The basic idea is to train a language model based on the manually labeled data, and then use the noisy emoticon data for smoothing. Experiments on real data sets demonstrate that ESLAM can effectively integrate both kinds of data to outperform those methods using only one of them.

NeurIPS Conference 2012 Conference Paper

Isotropic Hashing

• Weihao Kong
• Wu-Jun Li

Most existing hashing methods adopt some projection functions to project the original data into several dimensions of real values, and then each of these projected dimensions is quantized into one bit (zero or one) by thresholding. Typically, the variances of different projected dimensions are different for existing projection functions such as principal component analysis (PCA). Using the same number of bits for different projected dimensions is unreasonable because larger-variance dimensions will carry more information. Although this viewpoint has been widely accepted by many researchers, it is still not verified by either theory or experiment because no methods have been proposed to find a projection with equal variances for different dimensions. In this paper, we propose a novel method, called isotropic hashing (IsoHash), to learn projection functions which can produce projected dimensions with isotropic variances (equal variances). Experimental results on real data sets show that IsoHash can outperform its counterpart with different variances for different dimensions, which verifies the viewpoint that projections with isotropic variances will be better than those with anisotropic variances.

AAAI Conference 2012 Conference Paper

Sparse Probabilistic Relational Projection

• Wu-Jun Li
• Dit-Yan Yeung

Probabilistic relational PCA (PRPCA) can learn a projection matrix to perform dimensionality reduction for relational data. However, the results learned by PRPCA lack interpretability because each principal component is a linear combination of all the original variables. In this paper, we propose a novel model, called sparse probabilistic relational projection (SPRP), to learn a sparse projection matrix for relational dimensionality reduction. The sparsity in SPRP is achieved by imposing on the projection matrix a sparsity-inducing prior such as the Laplace prior or Jeffreys prior. We propose an expectation-maximization (EM) algorithm to learn the parameters of SPRP. Compared with PRPCA, the sparsity in SPRP not only makes the results more interpretable but also makes the projection operation much more efficient without compromising its accuracy. All these are verified by experiments conducted on several real applications.

IJCAI Conference 2011 Conference Paper

Generalized Latent Factor Models for Social Network Analysis

• Wu-Jun Li
• Dit-Yan Yeung
• Zhihua Zhang

Homophily and stochastic equivalence are two primary features of interest in social networks. Recently, the multiplicative latent factor model (MLFM) is proposed to model social networks with directed links. Although MLFM can capture stochastic equivalence, it cannot model well homophily in networks. However, many real-world networks exhibit homophily or both homophily and stochastic equivalence, and hence the network structure of these networks cannot be modeled well by MLFM. In this paper, we propose a novel model, called generalized latent factor model (GLFM), for social network analysis by enhancing homophily modeling in MLFM. We devise a minorization-maximization (MM) algorithm with linear-time complexity and convergence guarantee to learn the model parameters. Extensive experiments on some real-world networks show that GLFM can effectively model homophily to dramatically outperform state-of-the-art methods.

AAAI Conference 2011 Conference Paper

Social Relations Model for Collaborative Filtering

• Wu-Jun Li
• Dit-Yan Yeung

We propose a novel probabilistic model for collaborative filtering (CF), called SRMCoFi, which seamlessly integrates both linear and bilinear random effects into a principled framework. The formulation of SRMCoFi is supported by both social psychological experiments and statistical theories. Not only can many existing CF methods be seen as special cases of SRMCoFi, but it also integrates their advantages while simultaneously overcoming their disadvantages. The solid theoretical foundation of SRMCoFi is further supported by promising empirical results obtained in extensive experiments using real CF data sets on movie ratings.

AAAI Conference 2010 Conference Paper

Gaussian Process Latent Random Field

• Guoqiang Zhong
• Wu-Jun Li
• Dit-Yan Yeung
• Xinwen Hou
• Cheng-Lin Liu

The Gaussian process latent variable model (GPLVM) is an unsupervised probabilistic model for nonlinear dimensionality reduction. A supervised extension, called discriminative GPLVM (DGPLVM), incorporates supervisory information into GPLVM to enhance the classification performance. However, its limitation of the latent space dimensionality to at most C − 1 (C is the number of classes) leads to unsatisfactorily performance when the intrinsic dimensionality of the application is higher than C − 1. In this paper, we propose a novel supervised extension of GPLVM, called Gaussian process latent random field (GPLRF), by enforcing the latent variables to be a Gaussian Markov random field with respect to a graph constructed from the supervisory information. In GPLRF, the dimensionality of the latent space is no longer restricted to at most C − 1. This makes GPLRF much more flexible than DGPLVM in applications. Experiments conducted on both synthetic and real-world data sets demonstrate that GPLRF performs comparably with DGPLVM and other state-ofthe-art methods on data sets with intrinsic dimensionality at most C − 1, and dramatically outperforms DG- PLVM on data sets when the intrinsic dimensionality exceeds C − 1.

IJCAI Conference 2009 Conference Paper

• Wu-Jun Li
• Dit-Yan Yeung

In many applications, the data, such as web pages and research papers, contain relation (link) structure among entities in addition to textual content information. Matrix factorization (MF) methods, such as latent semantic indexing (LSI), have been successfully used to map either content information or relation information into a lower-dimensional latent space for subsequent processing. However, how to simultaneously model both the relation information and the content information effectively with an MF framework is still an open research problem. In this paper, we propose a novel MF method called relation regularized matrix factorization (RRMF) for relational data analysis. By using relation information to regularize the content MF procedure, RRMF seamlessly integrates both the relation information and the content information into a principled framework. We propose a linear-time learning algorithm with convergence guarantee to learn the parameters of RRMF. Extensive experiments on real data sets show that RRMF can achieve state-of-the-art performance.

NeurIPS Conference 2009 Conference Paper

Probabilistic Relational PCA

• Wu-Jun Li
• Dit-Yan Yeung
• Zhihua Zhang

One crucial assumption made by both principal component analysis (PCA) and probabilistic PCA (PPCA) is that the instances are independent and identically distributed (i. i. d. ). However, this common i. i. d. assumption is unreasonable for relational data. In this paper, by explicitly modeling covariance between instances as derived from the relational information, we propose a novel probabilistic dimensionality reduction method, called probabilistic relational PCA (PRPCA), for relational data analysis. Although the i. i. d. assumption is no longer adopted in PRPCA, the learning algorithms for PRPCA can still be devised easily like those for PPCA which makes explicit use of the i. i. d. assumption. Experiments on real-world data sets show that PRPCA can effectively utilize the relational information to dramatically outperform PCA and achieve state-of-the-art performance.