Title :
Model-based reinforcement learning for a multi-player card game with partial observability
Author :
Fujita, Hajime ; Ishii, Shin
Author_Institution :
Nara Inst. of Sci. & Technol., Japan
Abstract :
This article presents a model-based reinforcement learning (RL) scheme for a card game, "Hearts". Since this is a large-scale multi-player game with partial observability, effective state estimation and optimal control based on an environmental model are required. In our method, the learning agent is controlled by a one-step-ahead utility prediction using opponent agents\´ models. The computational intractability is overcome by the sampling method over a specific subspace. Simulation results show that our model-based RL method can produce an agent comparable to a human expert for this realistic problem.
Keywords :
computer games; learning (artificial intelligence); multi-agent systems; environmental model; model-based reinforcement learning; multiplayer card game; one-step-ahead utility prediction; optimal control; partial observability; sampling method; state estimation; Computational modeling; Heart; Humans; Large-scale systems; Learning; Observability; Optimal control; Predictive models; Sampling methods; State estimation;
Conference_Titel :
Intelligent Agent Technology, IEEE/WIC/ACM International Conference on
Print_ISBN :
0-7695-2416-8
DOI :
10.1109/IAT.2005.99