A Bayesian Sampling Approach to Exploration in Reinforcement Learning
John AsmuthLihong LiMichael L. LittmanAli NouriDavid WingatePublished in: CoRR (2012)
Keyphrases
- reinforcement learning
- exploration exploitation
- active exploration
- active learning
- exploration strategy
- function approximation
- action selection
- bandit problems
- autonomous learning
- state space
- reinforcement learning algorithms
- model free
- model based reinforcement learning
- bayesian networks
- sequential monte carlo
- bayesian learning
- temporal difference
- learning problems
- multi armed bandit
- markov decision processes
- monte carlo
- learning algorithm
- dynamic programming
- supervised learning
- maximum likelihood
- sampling strategy
- bayesian estimation
- posterior distribution
- markov chain monte carlo
- function approximators
- sampled data
- sampling algorithm
- temporal difference learning
- unknown environments
- random sampling
- particle filter
- decision theory
- optimal policy
- learning process
- multi agent
- bayesian decision