SGD Finds then Tunes Features in Two-Layer Neural Networks with near-Optimal Sample Complexity: A Case Study in the XOR problem.
Margalit GlasgowPublished in: CoRR (2023)
Keyphrases
- sample complexity
- neural network
- feature extraction
- special case
- machine learning
- pac learning
- feature space
- vc dimension
- learning problems
- feature set
- irrelevant features
- training examples
- active learning
- feature vectors
- theoretical analysis
- classification accuracy
- learning theory
- generalization error
- prior knowledge
- learning algorithm
- data sets