COLT
2004
Springer
13 years 10 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 ...