Learning and Evaluation Algorithms in Games

OData support
Supervisor:
Rajacsics Tamás
Department of Automation and Applied Informatics

Because of their complexity, games represent an ideal environment for research on artificial intelligence and they challenge all AI areas. Games theory is the branch of mathematics focusing on the analysis of the games and provides the necessary framework to build an AI on a game.

As part of my master’s degree thesis work a research about game theory and its role in the relation between games and artificial intelligence was presented. A strategy board game called Ludo star was developed to apply AI on. Different algorithms reflecting strategies that can be used to play the game were created based on well-defined and efficient scoring functions. Non-player characters used the strategy algorithms developed to test their efficiency and to eventually improve the algorithms and come up with a better one called Only-pros.

Minimax which is the solution to zero-sum games was implemented on the game developed, it was tested against the other algorithms created and optimized. Results showed that the algorithms worked well on the game, following the minimax algorithm gives better results than the other algorithms, still the game is not solved because of the game nature which includes a chance attribute.

Downloads

Please sign in to download the files of this thesis.