I received my PhD in computer science from CMU CSD, co-advised by Zico Kolter and Pradeep Ravikumar. I study machine learning theory and algorithms. My primary interest are representation learning and generalization theory, including statistical generalization and out-of-distribution (OOD) generalization. My dissertation establishes the contexture theory, which is a new framework for characterizing the mechanism of representation learning. My theory shows that representations are learned from the association between the input and a context variable, and increasing the model size alone inevitably produces diminishing returns, so further improvement requires better contexts. I also demonstrate ways to obtain such better contexts.

I received my Bachelor's degree in computer science and applied math (double degree) from Peking University, where I was advised by Liwei Wang.

Peer review:

• JMLR
• TMLR
• Nature Communications
• ICLR 2023 - 2025
• NeurIPS 2022 - 2025
• ICML 2022 - 2025
• AISTATS 2023 - 2025
• KDD 2023 - 2025
• AAAI 2025
• CVPR 2025
• ICCV 2023 - 2025
• ECCV 2024
• ACCV 2024
• SDM 2024

Best reviewer: NeurIPS 2022, 2024; AISTATS 2025
Teaching:

• CMU 15-750: Algorithms in the Real World Fall 2024
• CMU 10-701: Introduction to Machine Learning Fall 2022 (Head TA)