Variance-Reduced Conservative Policy Iteration.
Naman AgarwalBrian BullinsKaran SinghPublished in: CoRR (2022)
Keyphrases
- policy iteration
- markov decision processes
- model free
- reinforcement learning
- least squares
- fixed point
- optimal policy
- policy evaluation
- sample path
- finite state
- variance reduction
- temporal difference
- markov decision process
- infinite horizon
- average reward
- convergence rate
- linear programming
- markov decision problems
- state space
- markov chain
- dynamic programming
- sufficient conditions
- average cost