Policy Iteration for Decentralized Control of Markov Decision Processes.
Daniel S. BernsteinChristopher AmatoEric A. HansenShlomo ZilbersteinPublished in: J. Artif. Intell. Res. (2009)
Keyphrases
- markov decision processes
- decentralized control
- policy iteration
- optimal policy
- state space
- finite state
- factored mdps
- dynamic programming
- policy evaluation
- average reward
- reinforcement learning
- transition matrices
- markov decision process
- infinite horizon
- markov decision problems
- partially observable
- reinforcement learning algorithms
- discounted reward
- markov games
- average cost
- reward function
- multi agent
- state and action spaces
- evolutionary algorithm