Minimizing Expected Termination Time in One-Counter Markov Decision Processes.
Tomás BrázdilAntonín KuceraPetr NovotnýDominik WojtczakPublished in: ICALP (2) (2012)
Keyphrases
- markov decision processes
- total reward
- state space
- optimal policy
- finite state
- policy iteration
- transition matrices
- reinforcement learning
- dynamic programming
- stationary policies
- reinforcement learning algorithms
- factored mdps
- average reward
- planning under uncertainty
- reachability analysis
- partially observable
- action space
- decision theoretic planning
- risk sensitive
- decision processes
- finite horizon
- markov decision process
- action sets
- infinite horizon
- semi markov decision processes
- data mining
- state and action spaces
- heuristic search
- model based reinforcement learning
- np hard
- interval estimation