Multiresolution State-Space Discretization Method for Q-Learning with Function Approximation and Policy Iteration.
Amanda Kathryn LamptonJohn ValasekPublished in: SMC (2009)
Keyphrases
- function approximation
- policy iteration
- reinforcement learning
- state space
- model free
- markov decision processes
- temporal difference
- reinforcement learning algorithms
- optimal policy
- markov decision process
- temporal difference learning
- policy evaluation
- markov chain
- average reward
- markov decision problems
- function approximators
- actor critic
- learning algorithm
- naive bayes classifier
- dynamic programming
- machine learning
- state variables
- infinite horizon
- action space
- reinforcement learning methods
- basis functions
- particle filter
- initial state
- bayesian networks
- partially observable
- optimal control
- active learning
- partially observable markov decision processes
- policy gradient
- finite state