Policy iteration type algorithms for recurrent state Markov decision processes.
Stephen D. PatekPublished in: Comput. Oper. Res. (2004)
Keyphrases
- policy iteration
- markov decision processes
- state space
- factored mdps
- model free
- optimal policy
- reinforcement learning
- sample path
- markov decision process
- fixed point
- finite state
- approximate dynamic programming
- policy evaluation
- discounted reward
- action space
- dynamic programming
- partially observable
- markov decision problems
- infinite horizon
- least squares
- transition matrices
- reinforcement learning algorithms
- temporal difference
- policy iteration algorithm
- markov games
- average reward
- finite horizon
- convergence rate
- stochastic games
- linear programming
- real time dynamic programming
- decision processes
- data mining
- function approximation