PAC Optimal Exploration in Continuous Space Markov Decision Processes.
Jason PazisRonald ParrPublished in: AAAI (2013)
Keyphrases
- markov decision processes
- continuous space
- dynamic programming
- model based reinforcement learning
- average cost
- average reward
- finite horizon
- optimal policy
- finite state
- state space
- action sets
- interval estimation
- discrete space
- reinforcement learning
- stationary policies
- planning under uncertainty
- policy iteration
- infinite horizon
- total reward
- markov decision process
- decision theoretic planning
- discounted reward
- optimal control
- reachability analysis
- transition matrices
- partially observable
- optimality criterion
- optimal solution
- mathematical morphology
- distance measure
- data mining