Professor
Institute for Interdisciplinary Information Sciences
Tsinghua University
Research direction: Secure and robust AI application; AI Security; Privacy-preserving applications; High-performance distributed LLM training and inference.
Introduction to the Research Group
Secure and efficient AI applications and are committed to exploring and addressing challenges in AI security, privacy protection, and large-scale distributed systems. Research areas are extensive, including but not limited to:
1. Secure and Robust AI Applications: We design and develop secure and reliable practical AI applications, deploying them in real-world scenarios such as healthcare institutions to promote the implementation of AI technologies.
2. AI Security: We investigate security issues faced by current large models, including but not limited to jailbreaking tools, factual errors, hallucinations in large models, and privacy concerns.
3. Privacy-Preserving Applications: We combine techniques such as secure multi-party computation and federated learning to design large-scale, efficient privacy-preserving applications.
4. High-Performance Distributed LLM Training and Inference: We strive to optimize the training and inference processes of large language models. Through innovative distributed architectures and efficient algorithm design, we significantly enhance performance and scalability.
Research Achievements
In recent years, the research group has achieved excellent research results in the above-mentioned four areas, publishing numerous high-level papers in top conferences such as ACL, ICLR, ASPLOS, NSDI, VLDB, and CCS.
Publications
