Agnostic Q-learning with Function Approximation in Deterministic Systems: Tight Bounds on Approximation Error and Sample Complexity.
Simon S. DuJason D. LeeGaurav MahajanRuosong WangPublished in: CoRR (2020)
Keyphrases
- function approximation
- approximation error
- sample complexity
- tight bounds
- reinforcement learning
- temporal difference learning algorithms
- upper bound
- model free
- learning tasks
- temporal difference
- learning algorithm
- td learning
- function approximators
- radial basis function
- learning problems
- theoretical analysis
- supervised learning
- special case
- active learning
- lower bound
- generalization error
- principal component analysis
- training data
- temporal difference methods
- feature selection