Chenghao Liu currently is a Ph.D. candidate at Department of Data Science, City University of Hong Kong, mentored by Prof. Minghua Chen. He received his bachelor's degree in Mathematical Sciences, Zhejiang University in 2022.
His research lies at the intersection of machine learning theory and optimization, focuing on the theoretical foundations of machine learning to better explain and enhance practical performance, and developing theoretically grounded methods for fast optimization.
Publications
†: Corresponding Author; *: Equal Contribution
Learning + Optimization
Fast Projection-Free Approach (without Optimization Oracle) for Optimization over Compact Convex Set
Chenghao Liu, Enming Liang†, Minghua Chen†
NeurIPS2025 Spotlight, San Diego, USA
Key Words: projection-free, reparametrization, convex optimization
Learning Theory
Characterizing ResNet's Universal Approximation Capability
Chenghao Liu, Enming Liang, Minghua Chen†
ICML2024, Vienna, Austria
Key Words: ResNet, universal approximation, optimal approximation
ReLU Network with Width d+O(1) Can Achieve Optimal Approximation Rate
Chenghao Liu, Minghua Chen†
ICML2024, Vienna, Austria
Key Words: narrow neural network, optimal approximation