Research
My research lies at the intersection of reinforcement learning, statistical modeling, simulation, and sequential decision-making. I am especially interested in work that joins sound mathematical ideas with reproducible computation and careful empirical evaluation.
This page gives a concise public record rather than a project-by-project technical description. My CV contains the fuller research and presentation history, while the linked papers and repositories contain the technical details that are ready to be public.
Publications
Combinatorial Game Theory and Reinforcement Learning in Cumulative Tic-Tac-Toe via Evaluation Functions
Kai Li and Wei Zhu. Stats, 9(2), Article 28, 2026. First author.
Article and DOI · Official implementation
RL-QESA: Reinforcement-Learning Quasi-Equilibrium Simulated Annealing
Ruichen Xu, Kai Li, Haochun Wang, Georgios Kementzidis, Wei Zhu, and Yuefan Deng. AI for Math Workshop at ICML 2025. Co-first author.
OpenReview paper · Publication page
Manuscripts in preparation
- Reinforcement Learning with Expanded Action Spaces in Blackjack: A Controlled Six-Algorithm Comparison. Co-first-author manuscript in preparation with Taejin Park, Lichun He, Jeffery Liu, and Wei Zhu.
- Enhancing Regional Sea Level Predictions: A Unified Structural Equation Modeling Approach. Coauthored manuscript in preparation.
Code and manuscript links for the blackjack study will be added when the author team is ready to make the work public. At that point, the paper and implementation can have separate repositories.
Research experience
Institute for Advanced Computational Science, Stony Brook University
In summer 2025, I conducted a literature-based technical investigation of reinforcement learning for climate-model parameterization and multiagent adaptation to sea-level rise. The work involved literature synthesis, mathematical formulation, and technical presentations rather than a completed software implementation.
Mathematical Biosciences Institute, The Ohio State University
As an undergraduate research assistant, I studied deterministic and stochastic susceptible–infected–recovered models for epidemic dynamics.
School of Mathematics, Sun Yat-sen University
As a summer research assistant, I reviewed methods and phenotype summary statistics from genome-wide association studies.