Improving a Case-Based Texas Hold'em Poker Bot

摘要

This paper describes recent research that aims to improve upon our use of case-based reasoning in Texas hold'em poker bot called CASPER. CASPER uses knowledge of previous poker scenarios to inform its betting decisions. CASPER improves upon previous case-based reasoning approaches to poker and is able to play evenly against the University of Alberta's Pokibots and Simbots, from which it initially acquired its case-bases and updates previously published research by showing that CASPER plays profitably against human online competitors for play money. The new research described here shows how CASPER has improved its play against human players by remembering and reusing its own game play history.