ETRI-Knowledge Sharing Plaform

ENGLISH

성과물

논문 검색
구분 SCI
연도 ~ 키워드

상세정보

학술대회 Mastering Fighting Game Using Deep Reinforcement Learning With Self-play
Cited 1 time in scopus Download 3 time Share share facebook twitter linkedin kakaostory
저자
김대욱, 박성윤, 양성일
발행일
202008
출처
Conference on Games (CoG) 2020, pp.1-8
DOI
https://dx.doi.org/10.1109/CoG47356.2020.9231639
협약과제
20IH3300, 메타 플레이 인식 기반 지능형 게임 서비스 플랫폼 개발, 양성일
초록
One-on-one fighting game has played a role as a bridge between board game and real-time simulation game in terms of research on game AI because it needs middle-level computation power with medium-size complexity. In this paper, we propose a method to create fighting game AI agent using deep reinforcement learning with self-play and Monte Carlo Tree Search (MCTS). We also analyze various reinforcement learning configuration such as changes on state vector, reward shaping, and opponent compositions with novel performance metric. Agent trained by the proposed method was evaluated against other AIs. The evaluation result shows that mixing MCTS and self-play in a 1:3 ratio makes it possible to overwhelm other AIs in the game with 94.4% win rate. The fully-trained agent understands the game mechanism so that it waits until being close to enemy and performs actions at the optimal timing.
키워드
fighting game AI, FightingICE, Monte Carlo Tree Search, reinforcement learning, self-play
KSP 제안 키워드
AI Agent, Computation power, Deep reinforcement learning, Fighting game AI, Monte carlo tree search, Optimal timing, Reinforcement Learning(RL), Size complexity, State vector, board games, performance metrics