Institutional Repository of Vilnius University
Skatinamasis mokymas stalo žaidimams spręsti /
Abstract
dc:descriptionThis thesis examines how machine learning algorithms are used to train board game agents to make decisions that are likely to lead to positive outcomes. An experimental environment is created to conduct reinforcement learning experiments with different types of game agents. The research aims to find the best learning and performance parameters based on the results. During the work, the performance of board game agents is improved, and an alternative method for agents to play is implemented — board game strategies. The randomness factor of the game is addressed, and it is determined whether it affects learning. Experiments evaluate the suitability of the Q-learning algorithm for the selected board game and provide recommendations for applying reinforcement learning in the experimental environment. The results of this work confirm the findings of other studies that the standard Q-learning algorithm is not capable of solving tasks with large or infinite state spaces.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2024
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Razma, Karolis,
Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- lit
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD210646855&prefLang=en_US