Computer poker with intelligent agents

OData support
Supervisor:
Dr. Dobrowiecki Tadeusz Pawel
Department of Measurement and Information Systems

Computer poker is a widely researched field of artificial intelligence. Because of it's properties, it is used as a testbed for experimenting with new algorithms and methods in a controlled environment with reproducible results.

A few years ago I also used to play poker in online poker rooms as a hobby, so the fact that a computer can play this game, and more efficiently than even a skilled human player can, caught my attention. In spite of the great amount of progress that has already been made, designing and developing a computer poker bot is still a complex task with lots of problems to solve.

During my work I researched existing papers about computer poker, then I applied them to design my own computer poker player. My bot uses a machine learning based opponent modelling method for predicting actions, and the Monte Carlo tree search algorithm for making decisions.

After finishing the implementation, benchmarked the bot with other existing bots in the PokerAcademy simulator program.

Downloads

Please sign in to download the files of this thesis.