The optimality equation and ε-optimal strategies in Markov games with average reward criterion.
Heinz-Uwe KüenleRonald SchurathPublished in: Math. Methods Oper. Res. (2003)
Keyphrases
- average reward
- optimal strategy
- optimality criterion
- markov decision processes
- optimal policy
- stochastic games
- multiagent reinforcement learning
- decision problems
- reinforcement learning algorithms
- reinforcement learning
- long run
- model free
- state space
- expected cost
- monte carlo
- dynamic programming
- finite state
- markov chain
- policy iteration
- expected utility
- infinite horizon
- partially observable markov decision processes
- mathematical models
- markov decision problems
- evaluation function
- lower bound
- multiagent systems