Temperature Regret Matching for Imperfect-Information Games

Enmin Zhao, Kai Li, Junliang Xing

Keywords: regret matching, counterfactual regret minimization, imperfect-information games, regret updating

Abstract: Counterfactual regret minimization (CFR) methods are effective for solving two player zero-sum extensive games with imperfect information. Regret matching (RM) plays a crucial role in CFR and its variants to approach Nash equilibrium. In this paper, we present Temperature Regret Matching (TRM), a novel RM algorithm that adopts a different strategy. Also, we consider not only the opponent's strategy under the current strategy but also the opponent's strategies of the several last iterations for updating the external regret of each iteration. Furthermore, we theoretically demonstrate that the update of TRM converges to Nash Equilibrium. Competitive results in imperfect-information games have verified its effectiveness and efficiency.

One-sentence Summary: A new regret matching method, a novel regret updating method.

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics

Reviewed Version (pdf): https://openreview.net/references/pdf?id=lpXk3ZyzfK