People

To Prospective Students

There remains a significant gap between what modern AI systems can do and what we truly understand about them. My research group aims to help close this gap from two directions:

  • Bottom-up: building theoretical foundations and first principles for AI.
  • Top-down: analyzing state-of-the-art methods and propose theory-inspired new algorithms.

To me, achieving concrete understanding takes precedence over tuning black-box models for improvements. Coming from a statistics background, I believe statistics — along with other theoretical disciplines — has much to contribute to this effort. A key step in this process is conducting solid observational studies and formulating novel problems and hypotheses. Well-formulated toy settings can serve as powerful bridges between theory and practice.

I am actively seeking highly motived PhD/MPhil students and research assistants to join my group. Ideal candidates will have a strong theoretical foundation (in statistics, computer science, math, etc.) and hands-on experience in AI. If you want to contribute to this effort, please send me an email with your CV.

Current Members:

PhD Students

Former Members:

Reseach Interns at Huawei Noah’s Ark Lab.

  • Yimeng Chen, worked on model ensemble for domain generalization (ICML 2023), while pursuing a PhD at the University of Chinese Academy of Sciences.

  • Paweł Piwek, worked on boundary complexity for ReLU classifiers (UAI 2023), while pursuing a PhD at the University of Oxford.

  • Yi Liu, worked on optimal latent space for generative modeling, while pursuing a BS at CUHK-Shenzhen.

  • Weijian Luo, worked on diffusion distillation (NeurIPS 2023), while pursuing a PhD at Peking University.

  • Jiajun Ma, worked on diffusion sampling acceleration (ICLR 2024) and classifier guidance (ICML 2024), while pursuing a PhD at HKUST-Guangzhou.

  • Xuantong Liu, worked on conditional generation via model inversion (ICML 2024), while pursuing a PhD at HKUST.

  • Jakub Wornbard, worked on the duality between prompts and model weights in Transformers, while pursuing a MS at the University of Cambridge.

  • Shuchen Xue, worked on diffusion sampling acceleration (ICML 2024, CVPR 2024) and discrete diffusion models, while pursuing a PhD at the University of Chinese Academy of Sciences.

  • Yihong Luo, worked on diffusion distillation (ICLR 2025, arXiv 2025) and controllable generation (arXiv 2025, arXiv 2025), while pursuing a PhD at HKUST.