Mike Zhou
mikezhou [at] seas [dot] upenn [dot] edu
I am a recent computer science and mathematics graduate from the University of Pennsylvania. I’m interested in reinforcement learning, deep learning, and arithmetic combinatorics. My current focus is on exploring architectures and methods that could enable AI systems to reason more effectively, and on using reinforcement learning as a way to shape adaptive and generalizable behaviors.
During my undergraduate years, I was fortunate to work with Professors Michael Kearns, Dan Roth, and Ryan Cotterell, all of whom uniquely shaped my interest in AI.
latest posts
| Oct 22, 2025 | The Gradient of Thought: Reasoning as a Natural Objective for Alignment |
|---|---|
| Apr 12, 2025 | Diophantine Approximation and Transcendental Numbers |
| Oct 21, 2024 | Hardy-Ramanujan with probabilistic method |
selected publications
- Can Transformers Learn n-gram Language Models?In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing , 2024