Shiliang Sun

Chaoyue Ding, Shiliang Sun, Jing Zhao

Multimodal time series (MTS) anomaly detection is crucial for maintaining the safety and stability of working devices (e.g., water treatment system and spacecraft), whose data are characterized by multivariate time series with diverse modalities. Although recent deep learning methods show great potential in anomaly detection, they do not explicitly capture spatial-temporal relationships between univariate time series of different modalities, resulting in more false negatives and false positives. In this paper, we propose a multimodal spatial-temporal graph attention network (MST-GAT) to tackle this problem. MST-GAT first employs a multimodal graph attention network (M-GAT) and a temporal convolution network to capture the spatial-temporal correlation in multimodal time series. Specifically, M-GAT uses a multi-head attention module and two relational attention modules (i.e., intra- and inter-modal attention) to model modal correlations explicitly. Furthermore, MST-GAT optimizes the reconstruction and prediction modules simultaneously. Experimental results on four multimodal benchmarks demonstrate that MST-GAT outperforms the state-of-the-art baselines. Further analysis indicates that MST-GAT strengthens the interpretability of detected anomalies by locating the most anomalous univariate time series.

Zhanglin Wu, Tengfei Song, Ning Xie et al.

The rapid advancement of large vision-language models (LVLMs) has significantly propelled applications in document understanding, particularly in optical character recognition (OCR) and multilingual translation. However, current evaluations of LVLMs, like the widely used OCRBench, mainly focus on verifying the correctness of their short-text responses and long-text responses with simple layout, while the evaluation of their ability to understand long texts with complex layout design is highly significant but largely overlooked. In this paper, we propose Menu OCR and Translation Benchmark (MOTBench), a specialized evaluation framework emphasizing the pivotal role of menu translation in cross-cultural communication. MOTBench requires LVLMs to accurately recognize and translate each dish, along with its price and unit items on a menu, providing a comprehensive assessment of their visual understanding and language processing capabilities. Our benchmark is comprised of a collection of Chinese and English menus, characterized by intricate layouts, a variety of fonts, and culturally specific elements across different languages, along with precise human annotations. Experiments show that our automatic evaluation results are highly consistent with professional human evaluation. We evaluate a range of publicly available state-of-the-art LVLMs, and through analyzing their output to identify the strengths and weaknesses in their performance, offering valuable insights to guide future advancements in LVLM development. MOTBench is available at https://github.com/gitwzl/MOTBench.

Junxin Lu, Shiliang Sun

Air pollution, a pressing global problem, threatens public health, environmental sustainability, and climate stability. Achieving accurate and scalable forecasting across spatially distributed monitoring stations is challenging due to intricate multi-pollutant interactions, evolving meteorological conditions, and region specific spatial heterogeneity. To address this challenge, we propose AirPCM, a novel deep spatiotemporal forecasting model that integrates multi-region, multi-pollutant dynamics with explicit meteorology-pollutant causality modeling. Unlike existing methods limited to single pollutants or localized regions, AirPCM employs a unified architecture to jointly capture cross-station spatial correlations, temporal auto-correlations, and meteorology-pollutant dynamic causality. This empowers fine-grained, interpretable multi-pollutant forecasting across varying geographic and temporal scales, including sudden pollution episodes. Extensive evaluations on multi-scale real-world datasets demonstrate that AirPCM consistently surpasses state-of-the-art baselines in both predictive accuracy and generalization capability. Moreover, the long-term forecasting capability of AirPCM provides actionable insights into future air quality trends and potential high-risk windows, offering timely support for evidence-based environmental governance and carbon mitigation planning.

Zhilin Lin, Shiliang Sun

Reinforcement learning (RL) is playing an increasingly important role in fields such as robotic control and autonomous driving. However, the gap between simulation and the real environment remains a major obstacle to the practical deployment of RL. Agents trained in simulators often struggle to maintain performance when transferred to real-world physical environments. In this paper, we propose a latent space based approach to analyze the impact of simulation on real-world policy improvement in model-based settings. As a natural extension of model-based methods, our approach enables an intuitive observation of the challenges faced by model-based methods in sim-to-real transfer. Experiments conducted in the MuJoCo environment evaluate the performance of our method in both measuring and mitigating the sim-to-real gap. The experiments also highlight the various challenges that remain in overcoming the sim-to-real gap, especially for model-based methods.

Chenyu Lu, Shiliang Sun, Jing Zhao et al.

Multimodal machine translation (MMT) seeks to address the challenges posed by linguistic polysemy and ambiguity in translation tasks by incorporating visual information. A key bottleneck in current MMT research is the effective utilization of visual data. Previous approaches have focused on extracting global or region-level image features and using attention or gating mechanisms for multimodal information fusion. However, these methods have not adequately tackled the issue of visual information redundancy in MMT, nor have they proposed effective solutions. In this paper, we introduce a novel approach--multimodal machine translation with visual Scene Graph Pruning (PSG), which leverages language scene graph information to guide the pruning of redundant nodes in visual scene graphs, thereby reducing noise in downstream translation tasks. Through extensive comparative experiments with state-of-the-art methods and ablation studies, we demonstrate the effectiveness of the PSG model. Our results also highlight the promising potential of visual information pruning in advancing the field of MMT.

Zhuang Yu, Shiliang Sun, Jing Zhao et al.

Multimodal Machine Translation (MMT) aims to improve translation quality by leveraging auxiliary modalities such as images alongside textual input. While recent advances in large-scale pre-trained language and vision models have significantly benefited unimodal natural language processing tasks, their effectiveness and role in MMT remain underexplored. In this work, we conduct a systematic study on the impact of pre-trained encoders and decoders in multimodal translation models. Specifically, we analyze how different training strategies, from training from scratch to using pre-trained and partially frozen components, affect translation performance under a unified MMT framework. Experiments are carried out on the Multi30K and CoMMuTE dataset across English-German and English-French translation tasks. Our results reveal that pre-training plays a crucial yet asymmetrical role in multimodal settings: pre-trained decoders consistently yield more fluent and accurate outputs, while pre-trained encoders show varied effects depending on the quality of visual-text alignment. Furthermore, we provide insights into the interplay between modality fusion and pre-trained components, offering guidance for future architecture design in multimodal translation systems.

Daoming Zong, Shiliang Sun

Extreme multi-label text classification (XMTC) aims to tag a text instance with the most relevant subset of labels from an extremely large label set. XMTC has attracted much recent attention due to massive label sets yielded by modern applications, such as news annotation and product recommendation. The main challenges of XMTC are the data scalability and sparsity, thereby leading to two issues: i) the intractability to scale to the extreme label setting, ii) the presence of long-tailed label distribution, implying that a large fraction of labels have few positive training instances. To overcome these problems, we propose GNN-XML, a scalable graph neural network framework tailored for XMTC problems. Specifically, we exploit label correlations via mining their co-occurrence patterns and build a label graph based on the correlation matrix. We then conduct the attributed graph clustering by performing graph convolution with a low-pass graph filter to jointly model label dependencies and label features, which induces semantic label clusters. We further propose a bilateral-branch graph isomorphism network to decouple representation learning and classifier learning for better modeling tail labels. Experimental results on multiple benchmark datasets show that GNN-XML significantly outperforms state-of-the-art methods while maintaining comparable prediction efficiency and model size.

Xuli Sun, Shiliang Sun

Recent work has highlighted the vulnerability of many deep machine learning models to adversarial examples. It attracts increasing attention to adversarial attacks, which can be used to evaluate the security and robustness of models before they are deployed. However, to our best knowledge, there is no specific research on the adversarial attacks for multi-view deep models. This paper proposes two multi-view attack strategies, two-stage attack (TSA) and end-to-end attack (ETEA). With the mild assumption that the single-view model on which the target multi-view model is based is known, we first propose the TSA strategy. The main idea of TSA is to attack the multi-view model with adversarial examples generated by attacking the associated single-view model, by which state-of-the-art single-view attack methods are directly extended to the multi-view scenario. Then we further propose the ETEA strategy when the multi-view model is provided publicly. The ETEA is applied to accomplish direct attacks on the target multi-view model, where we develop three effective multi-view attack methods. Finally, based on the fact that adversarial examples generalize well among different models, this paper takes the adversarial attack on the multi-view convolutional neural network as an example to validate that the effectiveness of the proposed multi-view attacks. Extensive experimental results demonstrate that our multi-view attack strategies are capable of attacking the multi-view deep models, and we additionally find that multi-view models are more robust than single-view models.

Ziang Dong, Liang Mao, Shiliang Sun

We propose a generative model for adversarial attack. The model generates subtle but predictive patterns from the input. To perform an attack, it replaces the patterns of the input with those generated based on examples from some other class. We demonstrate our model by attacking CNN on MNIST.

Shiliang Sun, Zehui Cao, Han Zhu et al.

Machine learning develops rapidly, which has made many theoretical breakthroughs and is widely applied in various fields. Optimization, as an important part of machine learning, has attracted much attention of researchers. With the exponential growth of data amount and the increase of model complexity, optimization methods in machine learning face more and more challenges. A lot of work on solving optimization problems or improving optimization methods in machine learning has been proposed successively. The systematic retrospect and summary of the optimization methods from the perspective of machine learning are of great significance, which can offer guidance for both developments of optimization and machine learning research. In this paper, we first describe the optimization problems in machine learning. Then, we introduce the principles and progresses of commonly used optimization methods. Next, we summarize the applications and developments of optimization methods in some popular machine learning fields. Finally, we explore and give some challenges and open problems for the optimization in machine learning.

Jing Zhao, Jingjing Fei, Shiliang Sun

In order to better model high-dimensional sequential data, we propose a collaborative multi-output Gaussian process dynamical system (CGPDS), which is a novel variant of GPDSs. The proposed model assumes that the output on each dimension is controlled by a shared global latent process and a private local latent process. Thus, the dependence among different dimensions of the sequences can be captured, and the unique characteristics of each dimension of the sequences can be maintained. For training models and making prediction, we introduce inducing points and adopt stochastic variational inference methods.

Minghao Gu, Shiliang Sun

The Hamiltonian Monte Carlo (HMC) sampling algorithm exploits Hamiltonian dynamics to construct efficient Markov Chain Monte Carlo (MCMC), which has become increasingly popular in machine learning and statistics. Since HMC uses the gradient information of the target distribution, it can explore the state space much more efficiently than the random-walk proposals. However, probabilistic inference involving multi-modal distributions is very difficult for standard HMC method, especially when the modes are far away from each other. Sampling algorithms are then often incapable of traveling across the places of low probability. In this paper, we propose a novel MCMC algorithm which aims to sample from multi-modal distributions effectively. The method improves Hamiltonian dynamics to reduce the autocorrelation of the samples and uses a variational distribution to explore the phase space and find new modes. A formal proof is provided which shows that the proposed method can converge to target distributions. Both synthetic and real datasets are used to evaluate its properties and performance. The experimental results verify the theory and show superior performance in multi-modal sampling.

Jingjing Fei, Shiliang Sun

Online anomaly detection of time-series data is an important and challenging task in machine learning. Gaussian processes (GPs) are powerful and flexible models for modeling time-series data. However, the high time complexity of GPs limits their applications in online anomaly detection. Attributed to some internal or external changes, concept drift usually occurs in time-series data, where the characteristics of data and meanings of abnormal behaviors alter over time. Online anomaly detection methods should have the ability to adapt to concept drift. Motivated by the above facts, this paper proposes the method of sparse Gaussian processes with Q-function (SGP-Q). The SGP-Q employs sparse Gaussian processes (SGPs) whose time complexity is lower than that of GPs, thus significantly speeding up online anomaly detection. By using Q-function properly, the SGP-Q can adapt to concept drift well. Moreover, the SGP-Q makes use of few abnormal data in the training data by its strategy of updating training data, resulting in more accurate sparse Gaussian process regression models and better anomaly detection results. We evaluate the SGP-Q on various artificial and real-world datasets. Experimental results validate the effectiveness of the SGP-Q.

Omar Rivasplata, Emilio Parrado-Hernandez, John Shawe-Taylor et al.

PAC-Bayes bounds have been proposed to get risk estimates based on a training sample. In this paper the PAC-Bayes approach is combined with stability of the hypothesis learned by a Hilbert space valued algorithm. The PAC-Bayes setting is used with a Gaussian prior centered at the expected output. Thus a novelty of our paper is using priors defined in terms of the data-generating distribution. Our main result estimates the risk of the randomized algorithm in terms of the hypothesis stability coefficients. We also provide a new bound for the SVM classifier, which is compared to other known bounds experimentally. Ours appears to be the first stability-based bound that evaluates to non-trivial values.

Rongqing Huang, Shiliang Sun

Kernel regression is a popular non-parametric fitting technique. It aims at learning a function which estimates the targets for test inputs as precise as possible. Generally, the function value for a test input is estimated by a weighted average of the surrounding training examples. The weights are typically computed by a distance-based kernel function and they strongly depend on the distances between examples. In this paper, we first review the latest developments of sparse metric learning and kernel regression. Then a novel kernel regression method involving sparse metric learning, which is called kernel regression with sparse metric learning (KR$\_$SML), is proposed. The sparse kernel regression model is established by enforcing a mixed $(2,1)$-norm regularization over the metric matrix. It learns a Mahalanobis distance metric by a gradient descent procedure, which can simultaneously conduct dimensionality reduction and lead to good prediction results. Our work is the first to combine kernel regression with sparse metric learning. To verify the effectiveness of the proposed method, it is evaluated on 19 data sets for regression. Furthermore, the new method is also applied to solving practical problems of forecasting short-term traffic flows. In the end, we compare the proposed method with other three related kernel regression methods on all test data sets under two criterions. Experimental results show that the proposed method is much more competitive.

Shiliang Sun, Rongqing Huang, Ya Gao

Traffic flow forecasting, especially the short-term case, is an important topic in intelligent transportation systems (ITS). This paper does a lot of research on network-scale modeling and forecasting of short-term traffic flows. Firstly, we propose the concepts of single-link and multi-link models of traffic flow forecasting. Secondly, we construct four prediction models by combining the two models with single-task learning and multi-task learning. The combination of the multi-link model and multi-task learning not only improves the experimental efficiency but also the prediction accuracy. Moreover, a new multi-link single-task approach that combines graphical lasso (GL) with neural network (NN) is proposed. GL provides a general methodology for solving problems involving lots of variables. Using L1 regularization, GL builds a sparse graphical model making use of the sparse inverse covariance matrix. In addition, Gaussian process regression (GPR) is a classic regression algorithm in Bayesian machine learning. Although there is wide research on GPR, there are few applications of GPR in traffic flow forecasting. In this paper, we apply GPR to traffic flow forecasting and show its potential. Through sufficient experiments, we compare all of the proposed approaches and make an overall assessment at last.

Qingjiu Zhang, Shiliang Sun

Artificial neural network (ANN) is a very useful tool in solving learning problems. Boosting the performances of ANN can be mainly concluded from two aspects: optimizing the architecture of ANN and normalizing the raw data for ANN. In this paper, a novel method which improves the effects of ANN by preprocessing the raw data is proposed. It totally leverages the fact that different features should play different roles. The raw data set is firstly preprocessed by principle component analysis (PCA), and then its principle components are weighted by their corresponding eigenvalues. Several aspects of analysis are carried out to analyze its theory and the applicable occasions. Three classification problems are launched by an active learning algorithm to verify the proposed method. From the empirical results, conclusion comes to the fact that the proposed method can significantly improve the performance of ANN.

Shiliang Sun, Changshui Zhang, Yi Zhang

A novel predictor for traffic flow forecasting, namely spatio-temporal Bayesian network predictor, is proposed. Unlike existing methods, our approach incorporates all the spatial and temporal information available in a transportation network to carry our traffic flow forecasting of the current site. The Pearson correlation coefficient is adopted to rank the input variables (traffic flows) for prediction, and the best-first strategy is employed to select a subset as the cause nodes of a Bayesian network. Given the derived cause nodes and the corresponding effect node in the spatio-temporal Bayesian network, a Gaussian Mixture Model is applied to describe the statistical relationship between the input and output. Finally, traffic flow forecasting is performed under the criterion of Minimum Mean Square Error (M.M.S.E.). Experimental results with the urban vehicular flow data of Beijing demonstrate the effectiveness of our presented spatio-temporal Bayesian network predictor.

Feng Jin, Shiliang Sun

Traditional neural network approaches for traffic flow forecasting are usually single task learning (STL) models, which do not take advantage of the information provided by related tasks. In contrast to STL, multitask learning (MTL) has the potential to improve generalization by transferring information in training signals of extra tasks. In this paper, MTL based neural networks are used for traffic flow forecasting. For neural network MTL, a backpropagation (BP) network is constructed by incorporating traffic flows at several contiguous time instants into an output layer. Nodes in the output layer can be seen as outputs of different but closely related STL tasks. Comprehensive experiments on urban vehicular traffic flow data and comparisons with STL show that MTL in BP neural networks is a promising and effective approach for traffic flow forecasting.

Guoqing Chao, Shiliang Sun, Jinbo Bi

With advances in information acquisition technologies, multi-view data become ubiquitous. Multi-view learning has thus become more and more popular in machine learning and data mining fields. Multi-view unsupervised or semi-supervised learning, such as co-training, co-regularization has gained considerable attention. Although recently, multi-view clustering (MVC) methods have been developed rapidly, there has not been a survey to summarize and analyze the current progress. Therefore, this paper reviews the common strategies for combining multiple views of data and based on this summary we propose a novel taxonomy of the MVC approaches. We further discuss the relationships between MVC and multi-view representation, ensemble clustering, multi-task clustering, multi-view supervised and semi-supervised learning. Several representative real-world applications are elaborated. To promote future development of MVC, we envision several open problems that may require further investigation and thorough examination.

Shiliang Sun, John Paisley, Qiuyang Liu

Dirichlet processes (DP) are widely applied in Bayesian nonparametric modeling. However, in their basic form they do not directly integrate dependency information among data arising from space and time. In this paper, we propose location dependent Dirichlet processes (LDDP) which incorporate nonparametric Gaussian processes in the DP modeling framework to model such dependencies. We develop the LDDP in the context of mixture modeling, and develop a mean field variational inference algorithm for this mixture model. The effectiveness of the proposed modeling framework is shown on an image segmentation task.

Qiuyang Liu, Shiliang Sun

Gaussian processes (GPs) have been proven to be powerful tools in various areas of machine learning. However, there are very few applications of GPs in the scenario of multi-view learning. In this paper, we present a new GP model for multi-view learning. Unlike existing methods, it combines multiple views by regularizing marginal likelihood with the consistency among the posterior distributions of latent functions from different views. Moreover, we give a general point selection scheme for multi-view learning and improve the proposed model by this criterion. Experimental results on multiple real world data sets have verified the effectiveness of the proposed model and witnessed the performance improvement through employing this novel point selection scheme.

Shiliang Sun, John Shawe-Taylor, Liang Mao

This paper presents eight PAC-Bayes bounds to analyze the generalization performance of multi-view classifiers. These bounds adopt data dependent Gaussian priors which emphasize classifiers with high view agreements. The center of the prior for the first two bounds is the origin, while the center of the prior for the third and fourth bounds is given by a data dependent vector. An important technique to obtain these bounds is two derived logarithmic determinant inequalities whose difference lies in whether the dimensionality of data is involved. The centers of the fifth and sixth bounds are calculated on a separate subset of the training set. The last two bounds use unlabeled data to represent view agreements and are thus applicable to semi-supervised multi-view learning. We evaluate all the presented multi-view PAC-Bayes bounds on benchmark data and compare them with previous single-view PAC-Bayes bounds. The usefulness and performance of the multi-view bounds are discussed.

Shiliang Sun

This paper presents a new model called infinite mixtures of multivariate Gaussian processes, which can be used to learn vector-valued functions and applied to multitask learning. As an extension of the single multivariate Gaussian process, the mixture model has the advantages of modeling multimodal data and alleviating the computationally cubic complexity of the multivariate Gaussian process. A Dirichlet process prior is adopted to allow the (possibly infinite) number of mixture components to be automatically inferred from training data, and Markov chain Monte Carlo sampling techniques are used for parameter and latent variable inference. Preliminary experimental results on multivariate regression show the feasibility of the proposed model.

Shiliang Sun

We propose a new approach, multi-view Laplacian support vector machines (SVMs), for semi-supervised learning under the multi-view scenario. It integrates manifold regularization and multi-view regularization into the usual formulation of SVMs and is a natural extension of SVMs from supervised learning to multi-view semi-supervised learning. The function optimization problem in a reproducing kernel Hilbert space is converted to an optimization in a finite-dimensional Euclidean space. After providing a theoretical bound for the generalization performance of the proposed method, we further give a formulation of the empirical Rademacher complexity which affects the bound significantly. From this bound and the empirical Rademacher complexity, we can gain insights into the roles played by different regularization terms to the generalization performance. Experimental results on synthetic and real-world data sets are presented, which validate the effectiveness of the proposed multi-view Laplacian SVMs approach.