Revisiting CFR+ and Alternating Updates

TL;DR

This paper revisits the CFR+ algorithm for imperfect information games, correcting a previous proof error and providing updated theoretical guarantees while demonstrating its empirical efficiency.

Contribution

It offers corrected theoretical bounds for CFR+ and confirms its empirical performance improvements over prior algorithms.

Findings

Corrected proof of CFR+ theoretical bounds

Empirical results showing CFR+ efficiency

Validation of CFR+ as a practical solution method

Abstract

The CFR+ algorithm for solving imperfect information games is a variant of the popular CFR algorithm, with faster empirical performance on a range of problems. It was introduced with a theoretical upper bound on solution error, but subsequent work showed an error in one step of the proof. We provide updated proofs to recover the original bound.