Scholar
Weiyang Liu
Google Scholar ID: DMjROf0AAAAJ
CUHK | Max Planck Institute for Intelligent Systems
Citations & Impact
Citations
10,900
H-index
31
i10-index
56
Publications
20
Co-authors
26
list available
Publications
Cited
0
Resume (English only)
Background
- Assistant Professor in Computer Science and Engineering at The Chinese University of Hong Kong, heading the Scalable Principles for Learning and Reasoning Lab (SphereLab).
- Affiliated researcher at the Max Planck Institute for Intelligent Systems.
- Primary research focus: principled modeling of inductive bias in learning algorithms, investigating how inductive bias affects generalization.
- Develops "light-yet-sweet" learning algorithms: (i) light—conceptually simple and easy to implement; (ii) sweet—intuitive with non-trivial theoretical guarantees.
- Long-standing interest in geometric invariance, symmetry, and structures as guiding principles for generalization.
- Recently rethinking inductive bias for foundation models, with deep interest in large language models and generative modeling across visual, textual, and physical domains.
- Current research focuses on: (i) principled algorithms for training and adapting foundation models; (ii) understanding how LLMs perform reasoning and eliciting it in verifiable scenarios (e.g., formal/math/symbolic reasoning).
- Guided by two principles: insight must precede application; everything should be made as simple as possible, but not simpler.
