No ratings yet
Taught by or worked with Xianglin? Your rating helps other students choose.
Xianglin Yang is a research fellow at the National University of Singapore (NUS) School of Computing. He received his Ph.D. from NUS in 2025 and his B.S. from Fudan University in 2020. His research focuses on building trustworthy and reliable AI systems, with interests in AI safety and robustness, including defending large language models against adversarial attacks, securing autonomous agents, and benchmarking vulnerabilities in multi-modal systems. He also works on interpretability and debugging of deep learning models through interactive visualization and explanation techniques, as well as efficient and scalable machine learning methods such as sharpness-aware minimization.
Yang has published papers in venues including ICML, ACL, EMNLP, ESEC/FSE, NeurIPS, IJCAI, USENIX Security, and AAAI. Notable works include papers on bias manipulation attacks on LLM judges, interactive debugging approaches for deep classifiers, and time-travelling visualization techniques for model training. He has received the Dean’s Graduate Research Excellence Award from NUS and the NUS SOC Research Achievement Award. Yang serves as a reviewer for conferences such as ICLR, ICML, NeurIPS, IJCAI, AAAI, and CVPR, and has held research intern positions at ByteDance and Lenovo.
Explore the NUS Zombie Agents paper revealing how attackers can create persistent malicious control over self-evolving LLM agents, with key findings, implications, and defenses for cybersecurity.
Photo by Antoine Dautry on Unsplash
Photo by Chris Liverani on Unsplash
Photo by Peter Burdon on Unsplash
Photo by Andrii Zhuk on Unsplash
Photo by EESOFUFFZICH on Unsplash
Photo by Google DeepMind on Unsplash
Photo by Gabriele Malaspina on Unsplash
