Thomas Kaul just finished his bachelor’s thesis. The vision for his bachelor’s thesis was to create a computer program or rather an agent that plays Texas Hold’em to beat a specific human player in 1 vs 1. For instance, the agent should exploit people’s risk aversion, tendency to bluff, etc. To this goal, Thomas applied machine learning to learn the different playing styles of different human players and extract their gaming knowledge, game sense, and strategies. When facing a human opponent, the agent will use all these player models and compute for each the hypothetical outcomes. After an evaluation, the best performing player model is used to make the decisions for the current game. With this approach, the computer program is able to adapt to changed strategies of the opponent and play, implicitly identify the opponents strategy and weakness and exploit it.
Besides a configurable recommender system server, one of the result of this work is the following architecture for a computer agent playing Texas Hold’em.
One important side note for all supervisors out there. Take the passion of your students (in this case the passion for Texas Hold’em) and combine it with your area of expertise. You will be amazed by what cool things can happen and the quality of the outcome.
Abstract of the bachelor’s thesis
Poker provides an environment of great potential with well-defined rules for the research field of artificial intelligence. The popular card game provides incomplete information about the game state, non-deterministic outcome and stochastic elements where the outcome does not appear until thousands of hands have been played. These circumstances can be compared to making decisions in the real world and make the research interesting for other applications beyond poker. A major theme of this thesis is the development of an agent for Texas Hold’em Poker Sit and Go tournaments that plays skillful poker. For decision making, our approach is based on a recommender system. We mimic the behavior and strategies of a human poker player with an artificial intelligence agent. In various simulation setups we show that our approach is evaluated superior to simple poker opponents.
Thomas Kaul: Building an Agent for Texas Hold’em Poker Based on a Recommender System, University of Zurich, Faculty of Economics, 2011. (Bachelor Thesis)
Editors: Harald C. Gall, Amancio Bouza