COLT
2004
Springer
14 years 5 months ago
2004
Springer
Abstract. We consider Reinforcement Learning for average reward zerosum stochastic games. We present and analyze two algorithms. The first is based on relative Q-learning and the ...