I’m a first-year PhD student at UC Berkeley EECS, advised by Prof. Somayeh Sojoudi. I got my BS in computer science from Peking University previously, working with Prof. Yisen Wang and Dr. Yifei Wang on self-supervised learning and Prof. Lin Yang (UCLA) on reinforcement learning. My current research interests are self-supervised learning, reinforcement learning and LLM reasoning. Please feel free to contact me if you are interested in collaboration!

My CV is here.

Email: kaiwen_hu [at] berkeley (dot) edu

Research Interests

• Reasoning in LLMs: Understanding the causes of reasoning failures in large language models and solving them via reliable and robust methods.
• Self-Supervised Learning: Revisiting the existing self-supervised learning paradigms and uncovering their underlying mechanisms theoretically.
• Reinforcement Learning: Developing scalable RL algorithms and applying them to applications like LLMs.

News

• 2025.08: I am excited to start my PhD at UC Berkeley.
• 2025.03: I have accepted the PhD offer from UC Berkeley EECS. Many thanks!
• 2025.01: Our paper on contrastive learning theory is accepted at ICLR 2025!
• 2024.09: Our paper on equivariant self-supervised learning theory is accepted at NeurIPS 2024!

Publications

Value Gradient Flow: Behavior-Regularized RL Without Regularization
Haoran Xu*, Kaiwen Hu*, Somayeh Sojoudi, Amy Zhang
Under review.

Projection Head is Secretly an Information Bottleneck
Zhuo Ouyang*, Kaiwen Hu*, Qi Zhang, Yifei Wang, Yisen Wang
In ICLR 2025.
PDF | Code

Understanding the Role of Equivariance in Self-supervised Learning
Yifei Wang*, Kaiwen Hu*, Sharut Gupta, Ziyu Ye, Yisen Wang, Stefanie Jegelka
In NeurIPS 2024.
PDF | Code