Artificial Intelligence (AI) Mastering Development

Computing best response for imperfect-information extensive-form game

I have implemented the Counterfactual Regret Minimization (CFR) algorithm in recursive form. See for instance Algorithm 1 of An Introduction to Counterfactual Regret Minimization by Neller and Lanctot.

In order to evaluate the resulting strategies, I need to compute a best response to each of them and the corresponding expected value of the game. What is the recursive algorithm for this computation?

Leave a Reply

Your email address will not be published. Required fields are marked *