An empirical study of policy convergence in Markov decision process value iteration.
Christopher W. ZobelWilliam T. SchererPublished in: Comput. Oper. Res. (2005)
Keyphrases
- markov decision process
- optimal policy
- markov decision processes
- state space
- stationary policies
- reinforcement learning
- policy iteration
- infinite horizon
- finite horizon
- initial state
- stochastic shortest path
- markov games
- dynamic programming
- state action
- average cost
- convergence rate
- transition probabilities
- reward function
- action space
- decision processes
- finite state
- decision problems
- mathematical model