Haoyu Wang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Haoyu Wang has a total of 65 co-authors. Here are the top co-authors by number of shared publications:
- Xihe Qiu
Shanghai University of Engineering Science
16 shared publications - Xiaoyu Tan
National University of Singapore
15 shared publications - Bin Chen
University of Shanghai for Science and Technology
6 shared publications - Tianliang Yao
Department of Electronic Engineering, Faculty of Engineering, The Chinese University of Hong Kong, China
6 shared publications - Yu-Jie Xiong
School of Electronic and Electrical Engineering, Shanghai University of Engineering Science, Shanghai, China
4 shared publications - Zhijun Fang
School of Computer Science and Technology, Donghua University, Shanghai, China
4 shared publications
Latest Publications
- 14 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Medical Robotics and Bionics
- Vol. 8 (1) , 112-126
Large Language Models Meet Medical Robotics: A Technical Review
The convergence of large language models (LLMs) with medical robotics heralds a paradigm shift, transforming robotic systems into intelligent, language-aware collaborators in healthcare. This technical review examines the integration of LLMs into medical robotics, highlighting their capacity to elevate precision, adaptability, and human-robot synergy across diverse clinical applications. By embedding embodied intelligence, LLMs enable robots to interpret complex instructions and execute contextually informed actions. This advancement enhances autonomy and interaction in physical environments. Recent developments demonstrate that these capabilities boost operational efficiency and drive user-focused innovation. Nevertheless, formidable challenges persist, including real-time processing constraints, data privacy vulnerabilities, and ethical considerations surrounding model transparency and bias in safety-critical contexts. This paper synthesizes state-of-the-art strategies such as fine-tuning, prompt engineering, and multimodal integration. It assesses their efficacy and limitations within the stringent demands of medical settings. Looking forward, LLMs are poised to evolve into sophisticated multimodal architectures that seamlessly fuse language with sensory data to unlock highly autonomous, healthcare-tailored robotic systems. This review provides a thorough synthesis of current progress, compelling researchers to address the divide between AI innovation and clinical realization. It paves the way for a new era of responsive, intelligent medical robotics.
- 14 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Medical Robotics and Bionics
- Vol. 8 (1) , 112-126
Large Language Models Meet Medical Robotics: A Technical Review
The convergence of large language models (LLMs) with medical robotics heralds a paradigm shift, transforming robotic systems into intelligent, language-aware collaborators in healthcare. This technical review examines the integration of LLMs into medical robotics, highlighting their capacity to elevate precision, adaptability, and human-robot synergy across diverse clinical applications. By embedding embodied intelligence, LLMs enable robots to interpret complex instructions and execute contextually informed actions. This advancement enhances autonomy and interaction in physical environments. Recent developments demonstrate that these capabilities boost operational efficiency and drive user-focused innovation. Nevertheless, formidable challenges persist, including real-time processing constraints, data privacy vulnerabilities, and ethical considerations surrounding model transparency and bias in safety-critical contexts. This paper synthesizes state-of-the-art strategies such as fine-tuning, prompt engineering, and multimodal integration. It assesses their efficacy and limitations within the stringent demands of medical settings. Looking forward, LLMs are poised to evolve into sophisticated multimodal architectures that seamlessly fuse language with sensory data to unlock highly autonomous, healthcare-tailored robotic systems. This review provides a thorough synthesis of current progress, compelling researchers to address the divide between AI innovation and clinical realization. It paves the way for a new era of responsive, intelligent medical robotics.
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computational Social Systems
- Vol. 13 (4) , 1-13
ReMALIS: Inference-Guided Intention Propagation for Multiagent Stochastic Task Coordination With Large Language Models in Complex Networks Domains
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computational Social Systems
- Vol. 13 (4) , 1-13
ReMALIS: Inference-Guided Intention Propagation for Multiagent Stochastic Task Coordination With Large Language Models in Complex Networks Domains
- Open Access
EEG-VLM: A Hierarchical Vision-Language Model with Multi-Level Feature Alignment and Visually Enhanced Language-Guided Reasoning for EEG Image-Based Sleep Stage Prediction
Sleep stage classification based on electroencephalography (EEG) is fundamental for assessing sleep quality and diagnosing sleep-related disorders. However, most traditional machine learning methods rely heavily on prior knowledge and handcrafted features, while existing deep learning models still struggle to jointly capture fine-grained time-frequency patterns and achieve clinical interpretability. Recently, vision-language models (VLMs) have made significant progress in the medical domain, yet their performance remains constrained when applied to physiological waveform data, especially EEG signals, due to their limited visual understanding and insufficient reasoning capability. To address these challenges, we propose EEG-VLM, a hierarchical vision-language framework that integrates multi-level feature alignment with visually enhanced language-guided reasoning for interpretable EEG-based sleep stage classification. Specifically, a specialized visual enhancement module constructs high-level visual tokens from intermediate-layer features to extract rich semantic representations of EEG images. These tokens are further aligned with low-level CLIP features through a multi-level alignment mechanism, enhancing the VLM's image-processing capability. In addition, a Chain-of-Thought (CoT) reasoning strategy decomposes complex medical inference into interpretable logical steps, effectively simulating expert-like decision-making. Experimental results demonstrate that the proposed method significantly improves both the accuracy and interpretability of VLMs in EEG-based sleep stage classification, showing promising potential for automated and explainable EEG analysis in clinical settings.
Open AccessEEG-VLM: A Hierarchical Vision-Language Model with Multi-Level Feature Alignment and Visually Enhanced Language-Guided Reasoning for EEG Image-Based Sleep Stage Prediction
Sleep stage classification based on electroencephalography (EEG) is fundamental for assessing sleep quality and diagnosing sleep-related disorders. However, most traditional machine learning methods rely heavily on prior knowledge and handcrafted features, while existing deep learning models still struggle to jointly capture fine-grained time-frequency patterns and achieve clinical interpretability. Recently, vision-language models (VLMs) have made significant progress in the medical domain, yet their performance remains constrained when applied to physiological waveform data, especially EEG signals, due to their limited visual understanding and insufficient reasoning capability. To address these challenges, we propose EEG-VLM, a hierarchical vision-language framework that integrates multi-level feature alignment with visually enhanced language-guided reasoning for interpretable EEG-based sleep stage classification. Specifically, a specialized visual enhancement module constructs high-level visual tokens from intermediate-layer features to extract rich semantic representations of EEG images. These tokens are further aligned with low-level CLIP features through a multi-level alignment mechanism, enhancing the VLM's image-processing capability. In addition, a Chain-of-Thought (CoT) reasoning strategy decomposes complex medical inference into interpretable logical steps, effectively simulating expert-like decision-making. Experimental results demonstrate that the proposed method significantly improves both the accuracy and interpretability of VLMs in EEG-based sleep stage classification, showing promising potential for automated and explainable EEG analysis in clinical settings.
- 21 November 2025
- Published by Elsevier in Computer Methods and Programs in Biomedicine
- Vol. 274, 109163
MDM-DTA: Message Passing Neural Network with molecular descriptors and Mixture of Experts for drug–target affinity prediction
- 21 November 2025
- Published by Elsevier in Computer Methods and Programs in Biomedicine
- Vol. 274, 109163
MDM-DTA: Message Passing Neural Network with molecular descriptors and Mixture of Experts for drug–target affinity prediction
- 31 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Cognitive and Developmental Systems
- Vol. PP (23798920) , 1-15
Learning Expressive Task Embeddings and Sample-Efficient Exploration for Context Shift Reduction in Offline Meta-Reinforcement Environment
Context shift reduction for offline meta-reinforcement learning (CSRO) has made notable progress in addressing the context shift problem arising from the distribution mismatch between training and testing data. However, limitations remain in capturing rich task representations and efficiently exploring them during meta-testing. We propose two key enhancements: First, replacing the simple mean context encoding with an expressive attention-based mechanism to selectively focus on relevant transitions and learn intricate task representations for improved inference. Second, introducing an adaptive weighted prior exploration strategy that constructs a belief over possible tasks and iteratively refines this belief by weighting simulated trajectory rollouts under each hypothesis, enabling sample-efficient adaptation. Unlike random exploration, this strategy accelerates task representation updates by leveraging trajectory simulations under each task hypothesis. Through comprehensive experiments on diverse robotic control benchmarks with varying rewards and dynamics, we demonstrate significant performance gains over the original CSRO and surpass prior offline meta-RL methods. Our enhanced model highlights the importance of principled representation learning and exploration techniques in mitigating context shifts using only offline data while minimizing sample complexity during fast adaptation to new tasks. Our code is available at https://anonymous.4open.science/r/metaCSRL-A9B2.
- 31 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Cognitive and Developmental Systems
- Vol. PP (23798920) , 1-15
Learning Expressive Task Embeddings and Sample-Efficient Exploration for Context Shift Reduction in Offline Meta-Reinforcement Environment
Context shift reduction for offline meta-reinforcement learning (CSRO) has made notable progress in addressing the context shift problem arising from the distribution mismatch between training and testing data. However, limitations remain in capturing rich task representations and efficiently exploring them during meta-testing. We propose two key enhancements: First, replacing the simple mean context encoding with an expressive attention-based mechanism to selectively focus on relevant transitions and learn intricate task representations for improved inference. Second, introducing an adaptive weighted prior exploration strategy that constructs a belief over possible tasks and iteratively refines this belief by weighting simulated trajectory rollouts under each hypothesis, enabling sample-efficient adaptation. Unlike random exploration, this strategy accelerates task representation updates by leveraging trajectory simulations under each task hypothesis. Through comprehensive experiments on diverse robotic control benchmarks with varying rewards and dynamics, we demonstrate significant performance gains over the original CSRO and surpass prior offline meta-RL methods. Our enhanced model highlights the importance of principled representation learning and exploration techniques in mitigating context shifts using only offline data while minimizing sample complexity during fast adaptation to new tasks. Our code is available at https://anonymous.4open.science/r/metaCSRL-A9B2.
- 6 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Geoscience and Remote Sensing Letters
- Vol. 22 (1545598X) , 1-5
Adaptive Electromagnetic Analysis via Non-Euclidean Manifold Learning for Atmospheric Precipitation Understanding
The advent of dual-polarization meteorological sensing systems has revolutionized our capacity to comprehend atmospheric precipitation dynamics through electromagnetic signal analysis. However, the intricate non-linear relationships within high-dimensional polarimetric signatures present formidable challenges in extracting actionable intelligence for meteorological multimedia applications. This manuscript presents HyperSpectral-M, a computational framework that enhances polarimetric signal interpretation through systematic manifold learning approaches in non-Euclidean spaces, enabling more precise analysis of complex atmospheric phenomena. The proposed HyperSpectral-M framework addresses the limitations of existing methods by incorporating two key innovations: a signal disentanglement mechanism and a physics-constrained reconstruction paradigm. The disentanglement mechanism employs quaternion-based geodesic flow mapping coupled with adaptive spectral decomposition to project polarimetric signatures onto lower-dimensional manifolds while preserving critical microphysical properties. This is augmented by a multi-scale differential geometry analyzer that captures intricate spatiotemporal correlations across varying atmospheric conditions. The reconstruction paradigm leverages adversarial manifold alignment with structured probabilistic inference to synthesize high-fidelity radar representations while maintaining electromagnetic consistency constraints. HyperSpectral-M demonstrates significant real-world impact on meteorological applications by improving precipitation nowcasting accuracy by 15-20% compared to operational methods, enabling more timely and accurate flood warnings. Field validation with emergency management agencies shows that reduces false alarm rates by 30-40% while increasing lead time for severe weather warnings by 15-30 minutes.
- 6 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Geoscience and Remote Sensing Letters
- Vol. 22 (1545598X) , 1-5
Adaptive Electromagnetic Analysis via Non-Euclidean Manifold Learning for Atmospheric Precipitation Understanding
The advent of dual-polarization meteorological sensing systems has revolutionized our capacity to comprehend atmospheric precipitation dynamics through electromagnetic signal analysis. However, the intricate non-linear relationships within high-dimensional polarimetric signatures present formidable challenges in extracting actionable intelligence for meteorological multimedia applications. This manuscript presents HyperSpectral-M, a computational framework that enhances polarimetric signal interpretation through systematic manifold learning approaches in non-Euclidean spaces, enabling more precise analysis of complex atmospheric phenomena. The proposed HyperSpectral-M framework addresses the limitations of existing methods by incorporating two key innovations: a signal disentanglement mechanism and a physics-constrained reconstruction paradigm. The disentanglement mechanism employs quaternion-based geodesic flow mapping coupled with adaptive spectral decomposition to project polarimetric signatures onto lower-dimensional manifolds while preserving critical microphysical properties. This is augmented by a multi-scale differential geometry analyzer that captures intricate spatiotemporal correlations across varying atmospheric conditions. The reconstruction paradigm leverages adversarial manifold alignment with structured probabilistic inference to synthesize high-fidelity radar representations while maintaining electromagnetic consistency constraints. HyperSpectral-M demonstrates significant real-world impact on meteorological applications by improving precipitation nowcasting accuracy by 15-20% compared to operational methods, enabling more timely and accurate flood warnings. Field validation with emergency management agencies shows that reduces false alarm rates by 30-40% while increasing lead time for severe weather warnings by 15-30 minutes.
- 22 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Network Science and Engineering
- Vol. 13, 1084-1096
ROFED-LLM: Robust Federated Learning for Large Language Models in Adversarial Wireless Environments
Large language models (LLMs) have made significant advances in the field of natural language processing (NLP). However, their centralized training approach faces challenges related to data privacy, communication efficiency, and robustness against adversarial attacks, particularly in wireless environments. With the gradual depletion of high-quality public data, there is an urgent need to leverage private data distributed across various parties. Although federated learning (FL) offers a privacy-preserving collaborative training paradigm, it struggles to meet the high computational demands of edge devices and remains vulnerable to adversarial attacks. This paper introduces RoFed-LLM, a novel framework for robust, privacy-preserving training of LLMs on decentralized private data over wireless networks. By integrating split federated learning, which partitions the model across devices to enhance privacy, with adaptive jamming defense mechanisms, RoFed-LLM enables collaborative LLM training without raw data sharing while ensuring resilience against wireless adversarial attacks. Our multi-modal defense strategy combines model-level protections, such as differential privacy and dynamic pruning, with communication-level safeguards, including adaptive beamforming which optimizes wireless signal transmission to mitigate interference, and resource allocation optimization. Extensive experiments across diverse NLP tasks demonstrate RoFed-LLM’s superiority, achieving a 12.87% improvement in privacy preservation and 18.26% enhancement in jamming resilience compared to existing methods such as FedAvg and SCAFFOLD, with only a marginal 3.94% trade-off in model accuracy.
- 22 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Network Science and Engineering
- Vol. 13, 1084-1096
ROFED-LLM: Robust Federated Learning for Large Language Models in Adversarial Wireless Environments
Large language models (LLMs) have made significant advances in the field of natural language processing (NLP). However, their centralized training approach faces challenges related to data privacy, communication efficiency, and robustness against adversarial attacks, particularly in wireless environments. With the gradual depletion of high-quality public data, there is an urgent need to leverage private data distributed across various parties. Although federated learning (FL) offers a privacy-preserving collaborative training paradigm, it struggles to meet the high computational demands of edge devices and remains vulnerable to adversarial attacks. This paper introduces RoFed-LLM, a novel framework for robust, privacy-preserving training of LLMs on decentralized private data over wireless networks. By integrating split federated learning, which partitions the model across devices to enhance privacy, with adaptive jamming defense mechanisms, RoFed-LLM enables collaborative LLM training without raw data sharing while ensuring resilience against wireless adversarial attacks. Our multi-modal defense strategy combines model-level protections, such as differential privacy and dynamic pruning, with communication-level safeguards, including adaptive beamforming which optimizes wireless signal transmission to mitigate interference, and resource allocation optimization. Extensive experiments across diverse NLP tasks demonstrate RoFed-LLM’s superiority, achieving a 12.87% improvement in privacy preservation and 18.26% enhancement in jamming resilience compared to existing methods such as FedAvg and SCAFFOLD, with only a marginal 3.94% trade-off in model accuracy.
- Open Access
- 20 July 2025
- Published by Association for Computing Machinery (ACM)
- p. 2584-2595
Struct-X: Enhancing the Reasoning Capabilities of Large Language Models in Structured Data Scenarios
- 20 July 2025
- Published by Association for Computing Machinery (ACM)
- p. 2584-2595
Open AccessStruct-X: Enhancing the Reasoning Capabilities of Large Language Models in Structured Data Scenarios