Variance Reduction Techniques for Gradient Estimates in Reinforcement Learning.
Evan GreensmithPeter L. BartlettJonathan BaxterPublished in: J. Mach. Learn. Res. (2004)
Keyphrases
- variance reduction
- gradient estimation
- policy gradient
- reinforcement learning
- importance sampling
- confidence intervals
- actor critic
- sample size
- monte carlo
- bias variance decomposition
- function approximation
- quasi monte carlo
- reinforcement learning algorithms
- learning algorithm
- naive bayes classifier
- trade off
- approximate inference
- markov chain
- upper bound
- state space
- active learning