Fall 2023 GRASP SFI: Samuel Sokota, Carnegie Mellon University, “Reinforcement Learning in Two-Player Zero-Sum Games”

This is a hybrid event with in-person attendance in Levine 307 and virtual attendance on Zoom.

ABSTRACT

I’ll discuss reinforcement learning in two-player zero-sum games. Historically, this area has lacked algorithms that perform well in large settings with imperfect information. I’ll describe a resolution for making self-play RL performant.