Optimizing Local Satisfaction of Long-Run Average Objectives in Markov Decision Processes.
David KlaskaAntonín KuceraVojtech KurVít MusilVojtech RehákPublished in: CoRR (2023)
Keyphrases
- average cost
- long run
- markov decision processes
- optimal policy
- average reward
- short run
- infinite horizon
- finite horizon
- discounted reward
- reinforcement learning
- finite state
- expected cost
- state space
- stationary policies
- initial state
- policy iteration
- queueing networks
- markov decision process
- decision theoretic planning
- planning under uncertainty
- action sets
- search space
- control policy
- asymptotically optimal
- partially observable
- finite number
- decision problems
- markov decision problems
- multistage
- sufficient conditions
- dynamic programming
- reachability analysis
- transition matrices