Sample Complexity and Overparameterization Bounds for Temporal-Difference Learning With Neural Network Approximation.
Semih CayciSiddhartha SatpathiNiao HeR. SrikantPublished in: IEEE Trans. Autom. Control. (2023)
Keyphrases
- sample complexity
- temporal difference learning
- approximate value iteration
- neural network
- vc dimension
- upper bound
- lower bound
- fixed point
- function approximation
- theoretical analysis
- reinforcement learning
- learning problems
- learning algorithm
- generalization error
- evaluation function
- pac learning
- game playing
- special case
- temporal difference
- active learning
- function approximators
- supervised learning
- sample size
- reinforcement learning algorithms
- markov decision process
- worst case
- artificial neural networks
- state space
- average case
- monte carlo
- learning tasks
- training examples
- radial basis function
- dynamic programming
- optimal solution
- training data
- data mining