Kai Zhou
kaizhou@polyu.edu.hk
Assistant Professor, Department of Computing
The Hong Kong Polytechnic University
PQ836, (852) 2766 7244
Hung Hom, Kowloon, Hong Kong, China
I am an assistant professor in the Department of Computing at The Hong Kong Polytechnic University. My research interest centers around Security, with emphasis on AI Security, Data Security and Privacy, Adversarial Machine Learning, and Adversarial Network Analysis. My general research goal is to make intelligent systems secure, robust, privacy-aware, and trustworthy.
Join our group: We are always looking for self-motivated Ph.D. student/Research Assistant/Post-doc to join our group. I'm also happy to work with masters or undergraduate students at PolyU. If you are interested, please send me an email.
Current Research
We are now actively working on the following topics:- Adversarial Robustness of Graph-based Anomaly Detection: lots of graph analytic tools (e.g., social network analysis tools, GNNs, etc.) are used for anomaly detection. We study how to attack these tools and further develop defense approaches to make them robust.
- Security of Signed Graph Analysis: we study the security issues of analytic tasks over a specifical type of graphs termed signed graphs.
- Data Security and Privacy in Distributed Learning: while the distributed learning framework (e.g., Federated Learning) allows us to jointly learn from distributed data, there are important security issues, such as how to preserve user data privacy and how to ensure the learned machine learning model is robust and trustworthy under distributed attacks.
News
- [11/2021] Our work on structural attacks against graph-based anomaly detection is accepted for publication at ICDE'22. Congratuations to Yulin and Yuni.
- [09/2021] With our collaborator Prof. Xiapu Luo, one paper on structural attacks against Android malware detection is accepted to CCS'21.
- [09/2021] With our collaborator Prof. Tomasz P. Michalak, one paper on attacking sign prediction in signed graphs is accepted to ICDM'21.
- [08/2021] Yu Bu and Yuni Lai officially joined our group STiL as Ph.D. students. Welcome on board!
- [08/2021, Grant] Our project "Structural Attacks to Trust Analysis Systems in Signed Social Networks" is funded by the Young Scientist Fund, National Natural Science Foundation of China. ("针对符号社交互信网络分析系统的结构性攻击研究"，国家自然科学基金青年基金。)
- [06/2021, Grant] We are grateful that our project "Adversarial Robustness of Graph-based Anomaly Detection under Structural Attacks" is funded by the University Grants Committee (UGC) through the Early Career Scheme (ECS). A preliminary study BinarizedAttack: Structural Poisoning Attacks to Graph-based Anomaly Detection is released on arXiv.
- [06/2021, Grant] Our another project "Attacking Black-box Recommendations via User Profiles Generation under Hierarchical-structure Policy Gradient" is funded by the University Grants Committee (UGC) through the General Research Fund (GRF). I am happy to join this project as a Co-Investigator.
Academic Path
- Assistant Professor, Department of Computing at The Hong Kong Polytechnic University, 2020.7 - Present
- Post-doc, Vanderbilt University, 2018.5 - 2018.8, then moved to Washington University in St. Louis, 2018.8 - 2020.4
- Ph.D., Department of Electrical and Computer Engineering at Michigan State University, 2013.8 - 2018.5
- B.S., Shanghai Jiao Tong University, (电院), 2009.9 - 2013.7
Recent Community Service
Associate Editor- IET Communications, 2020 - Present
- 2022: AAAI, IJCAI, AAMSA, AAAI TRASE workshop
- 2021: IJCAI, AAMAS
