Applying Monte Carlo Tree Search with Deep Reinforcement Learning to Factorization Game
- 주제(키워드) Deep Q-Network , Monte Carlo Tree Search , Double DQN , Factorization Game
- 발행기관 서강대학교 일반대학원
- 지도교수 김종락
- 발행년도 2019
- 학위수여년월 2019. 2
- 학위명 석사
- 학과 및 전공 일반대학원 수학과
- 실제URI http://www.dcollection.net/handler/sogang/000000064079
- UCI I804:11029-000000064079
- 본문언어 영어
- 저작권 서강대학교 논문은 저작권보호를 받습니다.
초록/요약
Artificial intelligence, which is one of the core technologies of the fourth industrial revolution, is being used in the real life. In this thesis, we apply MCTS, Monte Carlo Tree Search, one of the reinforcement learning algorithms, to Factorization Game(matricking.com), which is a turn-based multiplayer board game. We used double DQN for rollout step. This is implemented in Python. As a result of the analysis, we achieved around 100% of the winning rate when we used Monte Carlo Tree Search with double DQN as a rollout strategy to board size n=4,5,6,7,8.
more