Yifeng Chu 褚毅锋

About

I am a final-year Ph.D. candidate in Electrical and Computer Engineering at the University of Illinois Urbana-Champaign, under the supervision of Maxim Raginsky.

Prior to my Ph.D., I also earned my M.S. in Electrical and Computer Engineering (2021) and my B.S. with high honors in Computer Engineering (2018) from UIUC.

Email: ychu26 [at] illinois [dot] edu.

Research

I am interested in the theoretical aspects of machine learning.

Much of my recent work focuses on Talagrand’s majorizing measure theorem and its various formulations and softmax extensions. This direction is motivated by the aim of better understanding the generalization abilities of modern machine learning models, in particular through frameworks that unify the geometric complexity of the hypothesis space with the information-theoretic stability of the learning algorithm.

Publications

Journal Articles

• Majorizing measures, codes, and information (extended version).
  Yifeng Chu and Maxim Raginsky.
  IEEE Transactions on Information Theory, in press. pdf

Conference Papers

• Talagrand meets Talagrand: upper and lower bounds on expected soft maxima of Gaussian processes with finite index sets.
  Yifeng Chu and Maxim Raginsky.
  Proceedings of the 37th International Conference on Algorithmic Learning Theory (ALT), 2026. Elegant Paper Award. arxiv
• A unified framework for information-theoretic generalization bounds.
  Yifeng Chu and Maxim Raginsky.
  Advances in Neural Information Processing Systems (NeurIPS), 2023. arxiv
• Majorizing measures, codes, information.
  Yifeng Chu and Maxim Raginsky.
  IEEE International Symposium on Information Theory (ISIT), 2023. arxiv

Preprints

• A chain rule for the expected suprema of Bernoulli processes.
  Yifeng Chu and Maxim Raginsky.
  arxiv, 2023.