Three-Subagent Adapting Architecture for Fighting Videogames
暂无分享,去创建一个
In standard fighting videogames, since opponents controlled by computers are in a rut, the user has learned their behaviors after long play and gets bored. Thus we propose an adapting opponent with three subagent architecture that adapts to the level of the user by reinforcement learning. The opponent was evaluated by human users by comparing it against static opponents.
[1] Ernest Adams,et al. Fundamentals of Game Design , 2006 .
[2] Richard S. Sutton,et al. Introduction to Reinforcement Learning , 1998 .
[3] Thore Graepel,et al. LEARNING TO FIGHT , 2004 .
[4] Richard S. Sutton,et al. Reinforcement Learning: An Introduction , 1998, IEEE Trans. Neural Networks.