Xiangrui Xu (许向蕊)        

I am now a third-year Ph.D Student in School of Computing Science from Beijing Jiaotong University since 2021, advised by Prof. Wei Wang. Before that, I received the B.A. and M.A. degrees in 2018 and 2021, respectively, in School of Mathematics and Computer Science, from Wuhan Polytechnic University.

My research mainly focuses on Trustworthy AI, especially privacy stealing attacks / defenses in distributed deep learning. My long-term goal is to advance the security and reliability of AI systems. Recently, I focus more on Trustworthy Large Foundation models (e.g., GPT and Diffusion Model), exploring their potential vulnerabilities and developing robust defense mechanisms to enhance their integrity and user trust.

News

• 1 paper was accepted to TIFS, 2025!.
• 1 paper cited more than 240 times on Google Scholar as of June 1, 2025!

Selected Publications [google scholar] [dblp]

1. VFLMonitor: Defending One-Party Hijacking Attacks in Vertical Federated Learning

   Xiangrui Xu, Zhize Li, Yufei Han, Bin Wang, Jiqiang Liu, Wei Wang*.

   IEEE Transactions on Information Forensics and Security (TIFS), 2025.

2. Finding the PISTE: Towards Understanding Privacy Leaks in Vertical Federated Learning Systems

   Xiangrui Xu, Wei Wang*, Zheng Chen, Bin Wang, Chao Li, Li Duan, Zhen Han, Yufei Han.

   IEEE Transactions on Dependable and Secure Computing (TDSC), 2024.

3. CGIR: Conditional Generative Instance Reconstruction Attacks Against Federated Learning

   Xiangrui Xu, Pengrui Liu, Wei Wang*, Hong-Liang Ma, Bin Wang, Zhen Han, Yufei Han.

   IEEE Transactions on Dependable and Secure Computing (TDSC), 2022.

4. Enhancing Privacy in Distributed Intelligent Vehicles With Information Bottleneck Theory

   Xiangrui Xu, Pengrui Liu, Yiwen Zhao, Lei Han, Wei Wang*, Yongsheng Zhu, Chongzhen Zhang, Bin Wang, Jian Shen, Zhen Han.

   IEEE Internet of Things Journal (IOTJ), 2024.

5. Threats, attacks and defenses to federated learning: issues, taxonomy and perspectives

   Pengrui Liu Xiangrui Xu (first co-author), Wei Wang*.

   Cybersecurity, 2022. Cited 179 times on Google Scholar as of January 1, 2025, Excellent Paper of Cybersecurity

6. AdvCat: Domain-Agnostic Robustness Assessment for Cybersecurity-Critical Applications with Categorical Inputs

   Helene Orsini; Hongyan Bao; Yujun Zhou; Xiangrui Xu; Yufei Han; Longyang Yi; Wei Wang; Xin Gao; Xiangliang Zhang*.

   2022 IEEE International Conference on Big Data (Big Data), 2022.

Patents

• Method For Distributing Equity Reward for Federated Learning Based on Equity Theory, Wei Wang, Xiangrui Xu, et al., US 2024/0046372A1. US Patent Granted

• Method for Privacy Leak detection method For Vertical Federated Learning Based on Feature Embedding analysis, Wei Wang, Xiangrui Xu, et al., CN116341004B, in Chinese. China Patent Granted

• Defense Method for Federated Learning Poisoning Attacks, Wei Wang, Xiangrui Xu, et al., CN116527393A, in Chinese. China Patent Granted

• Training Method for Conditional Generative Adversarial Networks, Yaqin Li, Xiangrui Xu, et al., CN111582348B, in Chinese. China Patent Granted

• Method for Robust Personalized Federated Learning, Wei Wang, Xiangrui Xu, et al., CN116962085B, in Chinese. China Patent Granted

• Method For Intelligent Detection for Federated Learning Based on Instance Reconstruction Attack, Wei Wang, Xiangrui Xu, et al., CN114548428B, in Chinese. China Patent Granted

• Defense Method for Federated Learning Poisoning Attacks, Wei Wang, Xiangrui Xu, et al., CN115907029B, in Chinese. China Patent Granted

• Method for Privacy Leak Detection for Federated Recommendation System, Wei Wang, Xiangrui Xu, et al., CN117592042A, in Chinese. China Patent Granted

• Method for Detecting and Identifying Abnormal Person Relationships Based on Federated Network, Wei Wang, Xiangrui Xu, et al., CN117010026B, in Chinese. China Patent Granted

Monographs

• Wang Wei et al., Privacy Preserving Computing[M], People's Posts and Telecommunications Press, 2023. Xiangrui Xu contributed to the writing of Section 9 (Machine Unlearning)

• Wang Wei et al., Artificial Intelligence Security Assessment Technology[M], Hans Publishing House, 2023. Xiangrui Xu served as an editorial board member

Research Projects

• Xu Xiangrui, Privacy Risk Assessment and Defense Methods of Federated Learning, innovation project for Ph.D. student at Beijing Jiaotong University, 2023-2025, in progress. PI

• Wang Wei, et al., Artificial Intelligence System Security Testing and Robust Enhancement Software and Hardware Integration, National Special Project, 2022-2024, in progress. Xiangrui Xu contributed to project application, management and R&D activities

• Wang Wei, et al., Trustworthy Mechanism and Key Technologies of Urban Intelligent Systems, National Key R&D Program Project, 2020-2023, in progress. Xiangrui Xu contributed to technology R&D

• Wang Wei, et al., Research on key technologies for security detection and protection of artificial intelligence applications, national special project, 2022-2024, in progress. Xiangrui Xu contributed to technology R&D

• Wang Wei, et al., Trustworthy distributed privacy-preserving machine learning methods and applications, enterprise cooperation project, 2022-2023, completed. Xiangrui Xu contributed to technology R&D

Selected Awards

• Awarded the National Scholarship during graduate studies; (2020)
• Received the First-Class Scholarship for two consecutive years of my Ph.D. studies. (2023-2024, 2024-2025)

Skills

• Research in Distributed AI Security: Established a solid research foundation in the field of trustworthy AI, with specific focus on enhancing security and privacy in distributed AI;
• Programming Proficiency: Skilled in programming languages such as Python, adept at developing efficient, scalable, and robust code;
• Data Analysis and Visualization Tools: Skilled in using software tools like Matlab, Office, etc., for data acquisition, quantitative analysis, and graphical representation.
• Proficient in English: Skilled in both oral and written communication.

Contact

Xiangrui Xu

Email: xiangrui.xu AT bjtu DOT eud DOT com