Improved bound on the worst case complexity of Policy Iteration.
Romain HollandersBalázs GerencsérJean-Charles DelvenneRaphaël M. JungersPublished in: CoRR (2014)
Keyphrases
- policy iteration
- markov decision processes
- model free
- sample path
- fixed point
- reinforcement learning
- least squares
- temporal difference
- upper bound
- optimal policy
- convergence rate
- markov decision process
- finite state
- average reward
- infinite horizon
- policy evaluation
- markov chain
- evaluation function
- model checking
- monte carlo