Author name cluster

Rui Tang

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

9 papers

1 author row

EAAI Journal 2026 Journal Article

A zero-shot prototype expansion model for alleviating the hubness problem and compound fault diagnosis

Lv Wang
Junyu Qi
Rui Tang
Qijun Wen
Yi Qin

Current zero-shot fault diagnosis methods employ a direct mapping strategy between the signal feature and attribute prototype, and use nearest neighbor estimation as the metric. This mapping-metric approach can lead to the hubness problem, i. e. , some samples of a prototype are recognized as other prototypes due to the nearest-neighbor estimation, affecting the diagnostic accuracy. To tackle this problem, a prototype expansion mapping method is constructed. A novel kernel function is proposed to expand prototype dimensions and increase the distance between prototypes, overcoming the hubness problem. Its ability to alleviate the hubness problem is verified through the intuitive illustration and theoretical analysis. Furthermore, since the existing metrics are not suitable for evaluating the hubness problem in zero-shot diagnosis scenarios, a new evaluation method is designed to evaluate the degree of the hubness problem. Moreover, a new attribute prototype definition approach is designed to mine the additional fault characteristics, enhancing zero-shot diagnostic capability. Building upon these innovations, we develop the zero-shot prototype expansion (ZSPE) model for compound fault diagnosis in rotating machinery. Experimental validation on bearing compound faults demonstrates ZSPE's ability to diagnose unseen compound faults without requiring any compound fault training samples, offering a solution to data acquisition challenges in real-world engineering contexts.

Details DOI

AAAI Conference 2026 Conference Paper

Autonomous Partner Selection for Cooperative Multi-Agent Reinforcement Learning

Rui Tang
Biao Luo
Yongzheng Cui

In cooperative Multi-Agent Reinforcement Learning (MARL), the subgroup-wise learning is employed to assign sub-tasks to agents towards the enhancement of team collaboration. However, the present work is dependent on manually defined allocation criteria, which hinders its capacity to adapt to environmental changes promptly, and also relaxes communication restrictions, thereby constraining the application of algorithms in a range of fields. In order to address these issues, the Autonomous Partner Selection (APS) framework is proposed, which offers an implicit grouping mechanism in an autonomous way. Each agent is capable of autonomously selecting cooperative partners and integrating their own observations with those of partners to harmonise the cooperative behaviour during the training stage. With a view to strictly restricting communication, the intention encoder is trained through information distillation, which enables agents to selectively take more cooperative actions based solely on local observations. Meanwhile, in order to circumvent potential conflicts engendered by homogenization behaviour, we employ a contrastive learning strategy to the cooperative intention generated by agents, thereby ensuring that the behavioural tendencies exhibited by different individuals remain as diverse as possible. Finally, extensive comparative experiments on the StarCraft Multi-Agent Challenge and Google Research Football are conducted. The results demonstrate that APS exhibits superior performance in comparison to the state-of-the-art algorithms across a range of tasks, and agents can adapt their grouping strategies in accordance with the environment to facilitate enhanced cooperation.

PDF Details DOI

AAAI Conference 2026 Conference Paper

Beyond Monotonicity: Revisiting Factorization Principles in Multi-Agent Q-Learning

Tianmeng Hu
Yongzheng Cui
Rui Tang
Biao Luo
Ke Li

Value decomposition is a central approach in multi-agent reinforcement learning (MARL), enabling centralized training with decentralized execution by factorizing the global value function into local values. To ensure individual-global-max (IGM) consistency, existing methods either enforce monotonicity constraints, which limit expressive power, or adopt softer surrogates at the cost of algorithmic complexity. In this work, we present a dynamical systems analysis of non-monotonic value decomposition, modeling learning dynamics as continuous-time gradient flow. We prove that, under approximately greedy exploration, all zero-loss equilibria violating IGM consistency are unstable saddle points, while only IGM-consistent solutions are stable attractors of the learning dynamics. Extensive experiments on both synthetic matrix games and challenging MARL benchmarks demonstrate that unconstrained, non-monotonic factorization reliably recovers IGM-optimal solutions and consistently outperforms monotonic baselines. Additionally, we investigate the influence of temporal-difference targets and exploration strategies, providing actionable insights for the design of future value-based MARL algorithms.

PDF Details DOI

AAAI Conference 2026 Conference Paper

Evolving Generalist Virtual Agents with Generative and Associative Memory

Zhenkui Zhang
Wendong Bu
Kaihang Pan
Bingchen Miao
Wenqiao Zhang
Guoming Wang
Wei Ji
Rui Tang

Generalist Virtual Agents (GVAs) powered by Multimodal Large Language Models (MLLMs) exhibit impressive capabilities. However, their long-term learning is hampered by a core limitation: a failure to evolve beyond existing trajectories. This stems from memory systems that treat experiences as isolated fragments and rely on brittle semantic retrieval, preventing the synthesis of novel solutions from disparate knowledge. To address this, we introduce CA3Mem, a framework inspired by the human hippocampus that organizes experiences into a structured memory graph. Leveraging this graph, CA3Mem features two key innovations: 1) a generative memory recombination mechanism that synthesizes novel solutions to drive agent evolution, and 2) an associative retrieval algorithm that employs spreading activation to recall a comprehensive and contextually-aware set of experiences. Experiments on OSWorld and WebArena demonstrate that CA3Mem significantly enhances agent capabilities, leading to marked improvements in long-horizon planning, compositional generalization for novel tasks, and continuous adaptation from experience.

PDF Details DOI

TAAS Journal 2025 Journal Article

Deceiving LLM through Compositional Instruction with Hidden Attacks

Shuyu Jiang
Xingshu Chen
Rui Tang

Recently, large language models (LLMs) have demonstrated promising applications in the autonomous driving (AD) domain, including language-based interactions and decision-making. Ensuring they safely handle harmful inputs is crucial before formal deployment. However, research reveals emerging hand-crafted jailbreak attacks, which pack harmful prompts into harmless instructions, can bypass LLMs’ security mechanisms and elicit harmful responses. To deeply understand such jailbreaks, this paper introduces a Compositional Instruction Attack (CIA) framework to generalize them, and develop two CIA jailbreaking methods to automatically generate tailored jailbreak prompts for each harmful prompt. Then, this paper builds the first CIA question-answering (CIAQA) dataset with 2.7K multiple-choice questions of 900 successful jailbreaks, for assessing LLMs’ ability to identify underlying harmful intents, harmfulness, and task priority in CIA jailbreaks. Combined with experimental analysis on CIAQA and other datasets, this paper concludes three possible reasons for the failure of LLM defenses against CIAs. Finally, we propose an intent-based defense paradigm (IBD), enabling LLMs to defend against CIA by leveraging its capability to identify intents. Experimental results show CIA can achieve attack success rates (ASR) of 95%+ and 85%+ in AD and common harmful scenarios for three well-known LLMs (GPT-4, GPT-3.5, and Llama2-70b-chat), and IBD reduces ASR by 74%+.

Details DOI

AAAI Conference 2025 Conference Paper

PrivDNFIS: Privacy-preserving and Efficient Deep Neuro-Fuzzy Inference System

Hao Ren
Xiao Lan
Rui Tang
Xingshu Chen

Deep Neuro-Fuzzy Inference Systems (DNFIS) seamlessly fuse neural networks with the fuzzy inference system enabling intricate decision-making and knowledge representation, while upholding a commendable degree of adaptability and interpretability. However, the challenge of privacy-preserving inference (PI) over DNFIS has remained largely uncharted, with no prior research addressing this critical issue. In this paper, we embark on an exploration of this issue. We introduce an efficient and secure PI framework for DNFIS, named PrivDNFIS, which leverages the post-quantum lattice-based homomorphic encryption to implement secure computation protocols for PI over DNFIS. Our work incorporates several non-trivial performance enhancements. Firstly, it consolidates multiple elements of input feature vectors into a single message, reducing encryption/decryption overhead. Secondly, building upon this novel encoding approach, PrivDNFIS can perform ciphertext aggregation and vector-vector inner production without necessitating time-consuming ciphertext rotation operations. Thirdly, we replace the softmax function in the DNFIS layer with a quadratic function to further enhance inference efficiency, without compromising the inference accuracy. Under the given threat model, we provide formal security proof for PrivDNFIS. In comprehensive experimental results, PrivDNFIS demonstrates an approximately 1.9 to 4.4 times reduction in end-to-end time cost compared to the benchmark.

PDF Details DOI

NeurIPS Conference 2025 Conference Paper

SpatialLM: Training Large Language Models for Structured Indoor Modeling

Yongsen Mao
Junhao Zhong
Chuan Fang
Jia Zheng
Rui Tang
Hao Zhu
Ping Tan
Zihan Zhou

SpatialLM is a large language model designed to process 3D point cloud data and generate structured 3D scene understanding outputs. These outputs include architectural elements like walls, doors, windows, and oriented object boxes with their semantic categories. Unlike previous methods which exploit task-specific network designs, our model adheres to the standard multimodal LLM architecture and is fine-tuned directly from open-source LLMs. To train SpatialLM, we collect a large-scale, high-quality synthetic dataset consisting of the point clouds of 12, 328 indoor scenes (54, 778 rooms) with ground-truth 3D annotations, and conduct a careful study on various modeling and training decisions. On public benchmarks, our model gives state-of-the-art performance in layout estimation and competitive results in 3D object detection. With that, we show a feasible path for enhancing the spatial understanding capabilities of modern LLMs for applications in augmented reality, embodied robotics, and more.

PDF Details

EAAI Journal 2024 Journal Article

TemporalHAN: Hierarchical attention-based heterogeneous temporal network embedding

Xian Mo
Binyuan Wan
Rui Tang

Heterogeneous temporal network embedding aims to learn each node of different types of a heterogeneous temporal network in each snapshot into a low-dimensional vector representation, which can be used for various network analysis tasks such as node classification and relationship prediction. Our work proposes a novel heterogeneous temporal graph neural network embedding framework (TemporalHAN) based on hierarchical attention using a temporal convolutional network (TCN). In particular, we introduce node-level and semantic-level attention into heterogeneous graph neural networks to identify the importance of different levels between nodes. For each snapshot, we first utilise a new random walk algorithm (NRWA) to collect strongly connected heterogeneous neighbours for each node of different types and group them by node types. In addition, the algorithm utilises a damping factor to ensure that the more recent snapshots allocate more random walk steps. We then utilise node-level and semantic-level attention to learn the importance between a node and its random walk neighbour for a specific node type and learn the importance of different node-type for this node, respectively. Finally, we adopt TCN to capture the evolution information between snapshots. Experimental results on relationship prediction and node classification reveal that the TemporalHAN is competitive against diverse state-of-the-art approaches. Our code is available at https: //github. com/Legendary-L/THAN.

Details DOI

YNICL Journal 2018 Journal Article

Neurotransmitter alterations in the anterior cingulate cortex in Crohn's disease patients with abdominal pain: A preliminary MR spectroscopy study

Kun Lv
Wenwen Song
Rui Tang
Zhiyong Pan
Yong Zhang
Yi Xu
Bin Lv
Yihong Fan

PURPOSE: H-MRS) to further explore the neural mechanism. METHODS: Sixteen CD patients with abdominal pain and 13 CD patients without abdominal pain, were recruited alongside 20 healthy controls (HCs) for this study. Clinical evaluations, including the 0-10 Visual Analogue Scale (VAS) of pain, Hospital Anxiety and Depression Scale (HADS) and Crohn's Disease Activity Index (CDAI), were evaluated prior to MR scanning. This study selected the bilateral ACC as the region of interest (ROI). The metabolites of the bilateral ACC were quantitatively analyzed by LCModel and Gannet. A independent sample t-test and one-way analysis of variance (ANOVA) were performed for statistical analysis. Spearman correlation analyses were performed to examine the relationship between the metabolite levels and clinical evaluations. RESULTS: The results indicated that CD patients with abdominal pain exhibited significantly higher levels of Glutamate (Glu)/(creatine + phosphocreatine, total creatine, tCr) over CD patients without abdominal pain, and HCs (p = 0.003, 0.009, respectively) in the bilateral ACC. The level of (Glutamate + Glutamine, Glx)/tCr of pain CD group was higher than non-pain CD group (p = 0.022). Moreover, within the pain CD group, Glu/tCr and Glx/tCr levels correlated strongly with the VAS scores of pain (ρ = 0.86, 0.59 respectively, p < 0.05). Meanwhile, the results indicates that CD patients with abdominal pain have significantly lower levels of γ-aminobutyric acid plus (GABA+)/tCr (p = 0.002) than HCs. To some extent, CDAI demonstrated a trend of negative correlation with GABA+/tCr levels (p = 0.088, ρ = -0.60). CONCLUSION: The neural mechanism of CD patients with abdominal pain in pain processing is tightly associated with neurochemical metabolites. An imbalance in Glu and GABA may play a key role in abdominal pain processing for patients with CD. This mechanism of pain may associate with the intestinal microbiota on the brain-gut axis.

Details DOI