Accelerated modified policy iteration algorithms for Markov decision processes.
Oleksandr ShlakhterChi-Guhn LeePublished in: Math. Methods Oper. Res. (2013)
Keyphrases
- policy iteration
- markov decision processes
- optimal policy
- reinforcement learning
- model free
- factored mdps
- policy evaluation
- sample path
- least squares
- state space
- fixed point
- average reward
- transition matrices
- finite state
- approximate dynamic programming
- markov decision process
- temporal difference
- dynamic programming
- finite horizon
- average cost
- planning under uncertainty
- partially observable
- convergence rate
- reinforcement learning algorithms
- decision processes
- markov decision problems
- actor critic
- action space
- stochastic games
- optimal control
- linear programming
- policy iteration algorithm
- stochastic shortest path