Institutional Repository of Vilnius University
7 Wonders žaidimo agentų kūrimo strategijos ir jų palyginimai /
Abstract
dc:descriptionThis thesis investigates the creation and comparative analysis of AI agents for the board game “7 wonders”. The study examines two main approaches: a deterministic method based on predefined rules and a Monte Carlo Tree Search (MCTS) algorithm, which uses simulation to predict outcomes and make decisions. The research focuses on enhancing MCTS by applying various strategies, such as reducing the set of moves based on domain knowledge and applying different simulation and move selection methods. The goal is to determine which MCTS strategies lead to the most effective agents. Using a simulator written in C#, extensive testing was conducted to evaluate the performance of these agents. The deterministic agent, while straightforward, lacked the adaptability inherent in MCTS. MCTSbased agents demonstrated superior results, particularly when integrated with various enhancing strategies. Techniques such as reducing search space and optimizing the move selection process were crucial in enhancing agents’ decisionmaking capabilities. The findings indicate that reducing the set of moves significantly improves the efficiency and performance of the MCTS algorithm. Additionally, the research highlights the importance of domainspecific knowledge in improving the MCTS process, allowing the agents to make more informed and strategic decisions. In conclusion, this work demonstrates the superiority of MCTS in developing advanced AI agents for complex strategic games. The study provides valuable insights into the strengths and weaknesses of different AI approaches, laying the foundation for future AI research and advancement in board games.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2024
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Bilevičius, Aleksandras,
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:ELABAETD210645189&prefLang=en_US