Weiyang Liu

Weiyang Liu is a machine learning researcher and incoming Assistant Professor in the Department of Computer Science and Engineering at The Chinese University of Hong Kong (CUHK), where he will lead the Scalable Principles for Learning and Reasoning Lab (SphereLab). He has previously held postdoctoral and research roles at the Max Planck Institute for Intelligent Systems, Georgia Tech, Google Brain, Nvidia Research, and MERL. His research focuses on principled modeling of inductive bias in learning algorithms, emphasizing geometric invariance, symmetry, and formal reasoning in large language models.

His work spans generative modeling (diffusion models, 3D mesh synthesis), foundation models (parameter-efficient training), and symbolic/mathematical reasoning evaluation. Publications include top-tier conferences like NeurIPS, ICML, CVPR, and ICLR, with Spotlight/Notable Top-25% designations.

He mentors PhD students including Zeju Qiu, Longhui Yu, and Zhouliang Yu, with alumni transitioning to institutions like Technical University of Munich, University of Tübingen, and Caltech. His lab emphasizes curiosity-driven research, theoretical rigor, and open collaboration.