Trick-taking poker game, as a popular form of imperfect information game, has been regarded as a challenge for a longtime . Since trick-taking game requires high level of not only reasoning, but also inference to excel . We train a strong GongzhuAI ScrofaZero from \textit{tabula rasa} by deep reinforcement learning . We introduce new techniques forimperfect information game including stratified sampling, importance weighting, .integral over equivalent class, Bayesian inference, etc. Our AI can achievehuman expert level performance. The methodologies in building our program can be easily transferred into a wide range of trick-tongzhu games. Our program can .be easily transfer into a range of tricks-taking games.

Author(s) : Naichen Shi, Ruichen Li, Sun Youran

Links : PDF - Abstract

Code :


Keywords : game - trick - program - learning - deep -

Leave a Reply

Your email address will not be published.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