Regularizing policy iteration for recursive feasibility and stability.
Mathieu GranzottoOlivier Lindamulage De SilvaRomain PostoyanDragan NesicZhong-Ping JiangPublished in: CDC (2022)
Keyphrases
- policy iteration
- markov decision processes
- optimal policy
- fixed point
- reinforcement learning
- model free
- least squares
- sample path
- markov decision process
- infinite horizon
- temporal difference
- average reward
- linear programming
- finite state
- markov decision problems
- state space
- actor critic
- policy evaluation
- long run
- optimal control
- sufficient conditions
- discounted reward
- convergence rate
- probabilistic model
- dynamic programming