我与南洋理工大学陶大程教授合作。 我的研究兴趣包括可信人工智能智能体、可解释人工智能和自动化科研。 我研究如何让人工智能智能体可靠、其决策可解释,以及人工智能系统如何自主开展科研,从文献分析、假设生成到实验设计与评估。 已发表工作主要研究机器学习系统的安全问题,尤其是基于机器学习的网络入侵检测中的攻击与防御。
我构建了 AI 科学家,一套可从零开始、全流程自动完成科研的系统:从建立知识库、调研文献、提出想法,到假设检验、实验、写作、审计乃至投稿。它要解答的问题是:在 AI 可以全自动科研的未来,人类在科研环节中的定位。
VulnerGAN: A Backdoor Attack through Vulnerability Amplification against Machine Learning-Based Network Intrusion Detection Systems Guangrui Liu, Weizhe Zhang, Xinjie Li, Kaisheng Fan, and Shui Yu Science China Information Sciences, 65(7):170303, 2022.
A Membership Inference and Adversarial Attack Defense Framework for Network Traffic Classifiers Guangrui Liu, Weizhe Zhang, Xurun Wang, Stephen King, and Shui Yu IEEE Transactions on Artificial Intelligence, 6(2):317–332, 2025.
Flow-MI: Reconstructing Private Traffic Patterns via Flow-Matching Model Inversion Kaisheng Fan, Guangrui Liu, Yishu Gao, Xinyue Sun, Jiahui Wu, Arun Kumar Sangaiah, and Weizhe Zhang IEEE Transactions on Consumer Electronics, 2026.
FMSA: A Meta-Learning Framework-Based Fast Model Stealing Attack Technique against Intelligent Network Intrusion Detection Systems Kaisheng Fan, Weizhe Zhang, Guangrui Liu, and Hui He Cybersecurity, 6(1):35, 2023.
面向智能网络入侵检测系统的基于边缘样本的数据污染防御方法 刘广睿, 张伟哲, 李欣洁 《计算机研究与发展》, 59(10):2348–2361, 2022.
Algal Bloom Trend Prediction Using Residual Attention Bidirectional LSTM Network K. Yang, Guangrui Liu, Y. Bao, and Y. Shi International Conference on Machine Learning and Computer Applications (ICMLCA), pp. 1–6, 2025.
Machine Learning-Based Detection of Gambling-Like Mobile Applications and Their Resource Coordination H. Su, Z. Ning, Z. Zhang, Z. Xi, C. Wu, Guangrui Liu, and Y. Cheng International Conference on Machine Learning, Big Data and Business Intelligence (MLBDBI), 2025.