Online Markov Decision Processes under Bandit Feedback.
Gergely NeuAndrás GyörgyCsaba SzepesváriAndrás AntosPublished in: NIPS (2010)
Keyphrases
- markov decision processes
- state space
- reinforcement learning
- policy iteration
- optimal policy
- dynamic programming
- finite state
- transition matrices
- reachability analysis
- state and action spaces
- partially observable
- online learning
- decision theoretic planning
- action space
- finite horizon
- markov decision process
- average reward
- planning under uncertainty
- average cost
- reinforcement learning algorithms
- infinite horizon
- model based reinforcement learning
- decision processes
- risk sensitive
- action sets
- data mining
- real valued
- interval estimation