Research
Model-Guided Public-Belief Search
Research tools for scalable search in imperfect-information games.
Central Question
Can learned models guide public-belief search without hiding strategic failure in imperfect-information games?
Main Contribution
This project develops research tools that connect learned models with public-belief search. A public belief gives the system a common state for reasoning when players hold different private information.
The repository emphasizes measurable search behavior, reproducible evaluation, and a clear boundary between learned guidance and strategic computation.
Citation
[1] J. Guo, “Model-Guided Public-Belief Search,” GitHub repository, 2026. [Online]. Available: https://github.com/jiaxing-guo/batr-research. [Accessed: Aug. 27, 2026].