Bounds for Synchronizing Markov Decision Processes.
Laurent DoyenMarie van den BogaardPublished in: CoRR (2022)
Keyphrases
- continuous state spaces
- markov decision processes
- state space
- action space
- discounted reward
- finite state
- reinforcement learning
- optimal policy
- upper bound
- transition matrices
- policy iteration
- lower bound
- dynamic programming
- finite horizon
- reachability analysis
- partially observable
- factored mdps
- infinite horizon
- average cost
- risk sensitive
- decision theoretic planning
- markov decision process
- action sets
- average reward
- planning under uncertainty
- upper and lower bounds
- state and action spaces
- variance reduction
- decision processes
- markov chain
- state abstraction
- objective function
- reward function
- model based reinforcement learning
- learning algorithm