Feng Hu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Feng Hu has a total of 50 co-authors. Here are the top co-authors by number of shared publications:
- Pengyue Li
School of Computer, Qinghai Normal University, Xining 810008, China
6 shared publications - Yurong Song
College of Automation, Nanjing University of Posts and Telecommunications, Nanjing 210023, China
4 shared publications - Zi-Ke Zhang
College of Media and International Culture, Zhejiang University, Hangzhou 310058, China
4 shared publications - Faxu Li
School of Computer, Qinghai Normal University , Xining 810008, Qinghai
3 shared publications Libing Bai
Department of Artificial Intelligence and Data Science, XinZhou Normal University
3 shared publications- You-Feng Zhao
College of Artificial Intelligence, Nanjing University of Posts and Telecommunications, Nanjing 210023, China
3 shared publications
Latest Publications
- Open Access
- 1 January 2027
- Published by Acta Physica Sinica, Chinese Physical Society and Institute of Physics, Chinese Academy of Sciences in Acta Physica Sinica
- Vol. 76 (7)
Opinion dynamics model based on temporal higher-order networks
揭示社交网络中的信息传播规律,是实现网络空间治理所面临的核心挑战。针对现有意见动力学模型难以还原信息传播过程中用户在线-离线动态演化以及高阶交互的局限,本文构建融合用户在线-离线状态切换机制的时序社交超网络,并在此基础上提出耦合“个体-圈层-群组”三层影响机制的连续意见动力学模型。通过理论分析,探究模型中立单稳态的局部渐进稳定边界条件,并揭示非对称分岔机理与多稳态涌现规律。通过在合成网络上的参数敏感性分析,以及在三类实证社交超网络上的实验验证,表明所提模型能够较好地捕捉舆情演化中的多稳态现象。研究结果为复杂社交环境下的信息监管提供理论支撑与治理启示。
- 1 January 2027
- Published by Acta Physica Sinica, Chinese Physical Society and Institute of Physics, Chinese Academy of Sciences in Acta Physica Sinica
- Vol. 76 (7)
Open AccessOpinion dynamics model based on temporal higher-order networks
揭示社交网络中的信息传播规律,是实现网络空间治理所面临的核心挑战。针对现有意见动力学模型难以还原信息传播过程中用户在线-离线动态演化以及高阶交互的局限,本文构建融合用户在线-离线状态切换机制的时序社交超网络,并在此基础上提出耦合“个体-圈层-群组”三层影响机制的连续意见动力学模型。通过理论分析,探究模型中立单稳态的局部渐进稳定边界条件,并揭示非对称分岔机理与多稳态涌现规律。通过在合成网络上的参数敏感性分析,以及在三类实证社交超网络上的实验验证,表明所提模型能够较好地捕捉舆情演化中的多稳态现象。研究结果为复杂社交环境下的信息监管提供理论支撑与治理启示。
- Open Access
- 28 October 2026
- Published by Frontiers Media SA in Frontiers in Medicine
- Vol. 13, 1974555
TSCNet: transformer-simplicial convolutional network for functional brain network classification
Functional brain network modeling provides an important tool for analyzing disease-related abnormalities in functional connectivity. Conventional graph models, however, treat nodes and edges as their basic objects and mainly encode pairwise relationships between brain regions, making it difficult to explicitly represent higher-order topological units jointly formed by multiple regions. We therefore propose TSCNet, a functional brain network classification framework that integrates simplicial convolution with a Transformer. TSCNet first maps multi-site-harmonized functional connectivity matrices to hierarchical simplicial complexes containing 0-, 1-, and 2-simplices. Hodge-Laplacian-driven simplicial convolutions then extract local topological features separately at each order. These order-specific features are projected into a shared representation space and passed to a Transformer to model global dependencies across simplices and brain regions. We evaluated TSCNet on three public fMRI datasets, ABIDE, ADHD-200, and Haxby, using stratified 10-fold cross-validation repeated five times. The complete model achieved classification accuracies of 81.23% on ABIDE and 82.42% on ADHD-200. Removing either the SCNN or Transformer component reduced performance, indicating that local higher-order topology and global information integration provide complementary information. Parameter-sensitivity analyses and topological-pattern visualizations further characterized the model’s responses to simplex order and construction threshold. TSCNet provides a framework for jointly analyzing local higher-order topology and whole-brain dependencies, although its generalization to unseen scanners and the clinical specificity of the identified higher-order patterns require validation in independent cohorts.
- 28 October 2026
- Published by Frontiers Media SA in Frontiers in Medicine
- Vol. 13, 1974555
Open AccessTSCNet: transformer-simplicial convolutional network for functional brain network classification
Functional brain network modeling provides an important tool for analyzing disease-related abnormalities in functional connectivity. Conventional graph models, however, treat nodes and edges as their basic objects and mainly encode pairwise relationships between brain regions, making it difficult to explicitly represent higher-order topological units jointly formed by multiple regions. We therefore propose TSCNet, a functional brain network classification framework that integrates simplicial convolution with a Transformer. TSCNet first maps multi-site-harmonized functional connectivity matrices to hierarchical simplicial complexes containing 0-, 1-, and 2-simplices. Hodge-Laplacian-driven simplicial convolutions then extract local topological features separately at each order. These order-specific features are projected into a shared representation space and passed to a Transformer to model global dependencies across simplices and brain regions. We evaluated TSCNet on three public fMRI datasets, ABIDE, ADHD-200, and Haxby, using stratified 10-fold cross-validation repeated five times. The complete model achieved classification accuracies of 81.23% on ABIDE and 82.42% on ADHD-200. Removing either the SCNN or Transformer component reduced performance, indicating that local higher-order topology and global information integration provide complementary information. Parameter-sensitivity analyses and topological-pattern visualizations further characterized the model’s responses to simplex order and construction threshold. TSCNet provides a framework for jointly analyzing local higher-order topology and whole-brain dependencies, although its generalization to unseen scanners and the clinical specificity of the identified higher-order patterns require validation in independent cohorts.
- 1 October 2026
- Published by Elsevier in Knowledge-Based Systems
- Vol. 353
HIC-DRL: A deep reinforcement learning approach for hypergraph influence maximization with independent cascade model
- 1 October 2026
- Published by Elsevier in Knowledge-Based Systems
- Vol. 353
HIC-DRL: A deep reinforcement learning approach for hypergraph influence maximization with independent cascade model
- 22 September 2026
- Published by Elsevier in Neurocomputing
- Vol. 708
hyperiDPath: A multilayer hypergraph drug repurposing path-reasoning framework based on hypergraph learning
- 22 September 2026
- Published by Elsevier in Neurocomputing
- Vol. 708
hyperiDPath: A multilayer hypergraph drug repurposing path-reasoning framework based on hypergraph learning
- Open Access
A Propagation Model of Social Hypernetwork Based on Directed Hypergraph
In the existing research on information propagation modeling in social networks, hypergraphs have been widely applied to characterize the high-order interaction relationships involving multiple nodes. However, most models are still based on the assumption of undirected connections, which leads to certain limitations in depicting the information flow direction and the structural characteristics of propagation chains. To address the above problems, a social hypernetwork propagation model with directional constraints is constructed in this paper by introducing the directed hypergraph structure and combining it with the improved SEIR model. The strength of social relationships is measured by intimacy in the model, and a comprehensive characterization of the information propagation process is achieved by integrating the threshold mechanism of the directed hypergraphs with the attenuation function of information timeliness. In addition, the effectiveness of the proposed model is verified by taking the event of “imposing additional tariffs” as an example, and the evolutionary characteristics of propagation in different network structures, as well as the impacts of user confidence and information timeliness, are analyzed using simulation experiments. The results indicate that the model is applicable to characterizing the information propagation trends and dynamic characteristics in real social networks, and can provide theoretical references and methodological support for the prediction and regulation of network public opinion.
Open AccessA Propagation Model of Social Hypernetwork Based on Directed Hypergraph
In the existing research on information propagation modeling in social networks, hypergraphs have been widely applied to characterize the high-order interaction relationships involving multiple nodes. However, most models are still based on the assumption of undirected connections, which leads to certain limitations in depicting the information flow direction and the structural characteristics of propagation chains. To address the above problems, a social hypernetwork propagation model with directional constraints is constructed in this paper by introducing the directed hypergraph structure and combining it with the improved SEIR model. The strength of social relationships is measured by intimacy in the model, and a comprehensive characterization of the information propagation process is achieved by integrating the threshold mechanism of the directed hypergraphs with the attenuation function of information timeliness. In addition, the effectiveness of the proposed model is verified by taking the event of “imposing additional tariffs” as an example, and the evolutionary characteristics of propagation in different network structures, as well as the impacts of user confidence and information timeliness, are analyzed using simulation experiments. The results indicate that the model is applicable to characterizing the information propagation trends and dynamic characteristics in real social networks, and can provide theoretical references and methodological support for the prediction and regulation of network public opinion.
- 8 January 2026
- Published by Elsevier in Computational Biology and Chemistry
- Vol. 122, 108880
A multi-layer hypergraph framework for drug-drug interaction prediction based on transformer and hypergraph convolution
- 8 January 2026
- Published by Elsevier in Computational Biology and Chemistry
- Vol. 122, 108880
A multi-layer hypergraph framework for drug-drug interaction prediction based on transformer and hypergraph convolution
- Open Access
- 12 December 2025
- Published by Association for Computing Machinery (ACM)
- p. 649-655
Research on Online and Offline Rumor Propagation Model and Control Strategy Based on Hypernetwork
The popularity of social media and the Internet has enabled rumors to spread more rapidly and extensively, causing severe harm to individuals, organizations, and society. Traditional single-layer network models struggle to capture the characteristics of rumor propagation through both online and offline channels in the digital age. To address this, this study analyzes the features of rumor dissemination across online and offline channels, divides the social system into an individual interaction layer and an online information dissemination layer, and achieves cross-layer dynamic coupling through hyperedge mapping. Based on the classic SEIR model, we introduce two propagation states-online and offline transmission-allowing nodes to exhibit asymmetric propagation capabilities in physical and virtual spaces. This leads to the proposal of an “individual-information” dual-layer hypernetwork coupled model named SEIIR (Susceptible-Exposed-Informed-online-Informed-offline-Recovered). Building on this model, we propose three intervention mechanisms across different dimensions, leveraging the authoritative release system of official media, the community guidance mechanism of opinion leaders, and intelligent identification and dynamic blocking technologies on online platforms. Cross-verification through simulations and real-world public opinion datasets was conducted. The results show that the model effectively captures the propagation characteristics of rumors, and all three proposed control strategies can reduce the impact of rumor spread and effectively contain its dissemination. This study, through the SEIIR model and its intervention strategies, provides theoretical support and practical pathways for rumor governance in real-world societies.
- 12 December 2025
- Published by Association for Computing Machinery (ACM)
- p. 649-655
Open AccessResearch on Online and Offline Rumor Propagation Model and Control Strategy Based on Hypernetwork
The popularity of social media and the Internet has enabled rumors to spread more rapidly and extensively, causing severe harm to individuals, organizations, and society. Traditional single-layer network models struggle to capture the characteristics of rumor propagation through both online and offline channels in the digital age. To address this, this study analyzes the features of rumor dissemination across online and offline channels, divides the social system into an individual interaction layer and an online information dissemination layer, and achieves cross-layer dynamic coupling through hyperedge mapping. Based on the classic SEIR model, we introduce two propagation states-online and offline transmission-allowing nodes to exhibit asymmetric propagation capabilities in physical and virtual spaces. This leads to the proposal of an “individual-information” dual-layer hypernetwork coupled model named SEIIR (Susceptible-Exposed-Informed-online-Informed-offline-Recovered). Building on this model, we propose three intervention mechanisms across different dimensions, leveraging the authoritative release system of official media, the community guidance mechanism of opinion leaders, and intelligent identification and dynamic blocking technologies on online platforms. Cross-verification through simulations and real-world public opinion datasets was conducted. The results show that the model effectively captures the propagation characteristics of rumors, and all three proposed control strategies can reduce the impact of rumor spread and effectively contain its dissemination. This study, through the SEIIR model and its intervention strategies, provides theoretical support and practical pathways for rumor governance in real-world societies.
- 12 December 2025
- Published by Elsevier in Expert Systems with Applications
- Vol. 304
FMGHA: Future momentum gradient-based attack on hypergraph neural networks
- 12 December 2025
- Published by Elsevier in Expert Systems with Applications
- Vol. 304
FMGHA: Future momentum gradient-based attack on hypergraph neural networks
- Open Access
- 27 August 2025
- Published by MDPI AG in Mathematics
- Vol. 13 (17) , 2751
Constrained Optimal Control of Information Diffusion in Online Social Hypernetworks
With the rapid development of online social networks, issues related to information security and public opinion control have increasingly attracted widespread attention. Therefore, this study establishes a constrained optimal control framework for information diffusion in online social networks, based on the (Susceptible Inactive–Susceptible Active–Exposed–Informed–Recovered) information diffusion model on social hypernetworks. This framework incorporates both cost and triggering constraints, with the goal of optimally regulating the information diffusion process through dynamic intervention strategies. The existence and uniqueness of the optimal solution are theoretically proven, and the corresponding optimal control strategy is derived. The effectiveness and generality of the model are demonstrated through experiments, and the impact of different combinations of control strategies on system performance enhancement is investigated. The results indicate that the proposed control framework can significantly improve system control effectiveness while satisfying all imposed constraints and exhibits strong generalizability. Not only does this study enrich the theoretical foundation of information diffusion control, but it also provides practical theoretical support for addressing real-world issues such as public opinion guidance and commercial marketing in online social networks.- 27 August 2025
- Published by MDPI AG in Mathematics
- Vol. 13 (17) , 2751
Open AccessConstrained Optimal Control of Information Diffusion in Online Social Hypernetworks
With the rapid development of online social networks, issues related to information security and public opinion control have increasingly attracted widespread attention. Therefore, this study establishes a constrained optimal control framework for information diffusion in online social networks, based on the (Susceptible Inactive–Susceptible Active–Exposed–Informed–Recovered) information diffusion model on social hypernetworks. This framework incorporates both cost and triggering constraints, with the goal of optimally regulating the information diffusion process through dynamic intervention strategies. The existence and uniqueness of the optimal solution are theoretically proven, and the corresponding optimal control strategy is derived. The effectiveness and generality of the model are demonstrated through experiments, and the impact of different combinations of control strategies on system performance enhancement is investigated. The results indicate that the proposed control framework can significantly improve system control effectiveness while satisfying all imposed constraints and exhibits strong generalizability. Not only does this study enrich the theoretical foundation of information diffusion control, but it also provides practical theoretical support for addressing real-world issues such as public opinion guidance and commercial marketing in online social networks. - 21 July 2025
- Published by Elsevier in Applied Mathematics and Computation
- Vol. 508
SEIAR rumor spreading model with antagonistic states in hypernetworks
- 21 July 2025
- Published by Elsevier in Applied Mathematics and Computation
- Vol. 508
SEIAR rumor spreading model with antagonistic states in hypernetworks