Bandit-Based Planning and Learning in Continuous-Action Markov Decision Processes.
Ari WeinsteinMichael L. LittmanPublished in: ICAPS (2012)
Keyphrases
- markov decision processes
- reinforcement learning
- partially observable
- state space
- stochastic games
- learning algorithm
- finite state
- reinforcement learning algorithms
- partially observable markov decision processes
- least squares
- heuristic search
- optimal policy
- policy iteration
- finite horizon
- action space
- dynamic programming
- blocks world
- planning under uncertainty