Towards Finite-Sample Convergence of Direct Reinforcement Learning.
Shiau Hong LimGerald DeJongPublished in: ECML (2005)
Keyphrases
- finite sample
- reinforcement learning
- sample size
- statistical learning theory
- uniform convergence
- error bounds
- nearest neighbor
- parzen window
- machine learning
- learning problems
- generalization bounds
- statistical learning
- learning algorithm
- neural network
- model selection
- reproducing kernel hilbert space
- sufficient conditions
- data sets
- worst case
- supervised learning
- learning theory
- supervised classification
- data dependent
- image registration
- semi supervised