Near Optimal Exploration-Exploitation in Non-Communicating Markov Decision Processes.
Ronan FruitMatteo PirottaAlessandro LazaricPublished in: CoRR (2018)
Keyphrases
- markov decision processes
- exploration exploitation
- reinforcement learning
- active learning
- bandit problems
- state space
- optimal policy
- finite state
- function approximation
- decision theoretic planning
- action sets
- decision problems
- transition matrices
- planning under uncertainty
- reachability analysis
- relevance feedback
- finite horizon
- average cost
- policy iteration
- action space
- multi agent
- learning algorithm
- average reward
- model based reinforcement learning
- multiple features
- model free
- infinite horizon
- machine learning
- optimal control
- transfer learning
- supervised learning
- partially observable
- dynamic programming
- learning process
- real time dynamic programming