Publications: Multiple papers accepted at ICML 2024, ICLR 2024, NeurIPS 2023, etc.; Research Projects: Optimization theories and accelerated algorithms for neural networks, catapult dynamics of SGD, effect of ReLU non-linear activation on NTK condition number, etc.
Research Experience
2021-2022: Research Scientist at Meta (formerly Facebook); 2022-2024: Postdoc at Halıcıoğlu Data Science Institute (HDSI), UC San Diego, working with Dr. Misha Belkin; Since 2024/08: Assistant Professor in the ECE department at Purdue University.
Education
Ph.D.: Computer Science, The Ohio State University (2021), advised by Dr. Misha Belkin; M.S. and B.S.: Physics, Tsinghua University.
Background
Research Interests: Theoretical foundation of deep learning and its applications, theoretical understanding of neural network models and training dynamics, solving practical problems. Research areas include: fundamental properties of neural networks, optimization theory, attention models, feature learning, etc.
Miscellany
Personal Interests: Experimentally finding new phenomena in deep learning and explaining them using mathematical tools, connections between optimization and generalization performance of neural networks.