Online Markov Decision Processes with Aggregate Bandit Feedback.
Alon CohenHaim KaplanTomer KorenYishay MansourPublished in: CoRR (2021)
Keyphrases
- markov decision processes
- finite state
- state space
- decision theoretic planning
- policy iteration
- dynamic programming
- online learning
- transition matrices
- optimal policy
- reinforcement learning
- reachability analysis
- partially observable
- state and action spaces
- factored mdps
- planning under uncertainty
- finite horizon
- decision processes
- markov decision process
- infinite horizon
- average cost
- action sets
- decision diagrams
- reinforcement learning algorithms
- model based reinforcement learning
- action space
- average reward
- sufficient conditions
- evaluation function