Tuesday, September 20, 2016 - 12:00pm to 1:00pm
Location:3305 Newell-Simon Hall
Speaker:NOAM BROWN, Ph.D. Student http://www.cs.cmu.edu/~noamb/
For More Information, Contact:firstname.lastname@example.org
Counterfactual Regret Minimization (CFR) is a leading algorithm for solving large zero-sum imperfect-information games. CFR is an iterative algorithm that repeatedly traverses the game tree, updating regrets at each information set. We introduce Regret-Based Pruning (RBP), an improvement to CFR that prunes any path of play in the tree, and its descendants, that has negative regret. It revisits that sequence at the earliest subsequent CFR iteration where the regret could have become positive, had that path been explored on every iteration. We prove that in zero-sum games it asymptotically prunes any action that is not part of a best response to some Nash equilibrium. This leads to provably faster convergence and lower space requirements. Experiments show that RBP can result in an order of magnitude reduction in both space and time needed to converge, and that the reduction factor increases with game size.
Presented in Partial Fulfillment of the CSD Speaking Skills Requirement.