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