On the Asymptotic Optimality of Finite Approximations to Markov Decision Processes with Borel Spaces.
Naci SaldiSerdar YükselTamás LinderPublished in: Math. Oper. Res. (2017)
Keyphrases
- markov decision processes
- asymptotic optimality
- state and action spaces
- asymptotically optimal
- optimal policy
- stationary policies
- policy evaluation
- sufficient conditions
- reinforcement learning
- decision theoretic planning
- finite state
- state space
- dynamic programming
- flowshop
- infinite horizon
- reinforcement learning algorithms
- partially observable
- action space
- planning under uncertainty
- policy iteration
- average cost
- reachability analysis
- decision processes
- transition matrices
- markov decision process
- average reward
- reward function
- finite number
- model based reinforcement learning
- arrival rate
- long run
- decision problems
- multistage
- scheduling problem
- search space
- learning algorithm