Reachability in Recursive Markov Decision Processes.
Tomás BrázdilVáclav BrozekVojtech ForejtAntonín KuceraPublished in: CONCUR (2006)
Keyphrases
- markov decision processes
- state space
- finite state
- optimal policy
- reinforcement learning
- decision theoretic planning
- transition matrices
- dynamic programming
- heuristic search
- reinforcement learning algorithms
- policy iteration
- partially observable
- reachability analysis
- planning under uncertainty
- markov chain
- infinite horizon
- finite horizon
- decision processes
- factored mdps
- action sets
- search space
- action space
- reward function
- state abstraction
- model based reinforcement learning
- dynamical systems
- interval estimation