Neural Tangent Kernel Eigenvalues Accurately Predict Generalization.
James B. SimonMadeline DickensMichael Robert DeWeesePublished in: CoRR (2021)
Keyphrases
- kernel matrix
- kernel function
- network architecture
- neural network
- gram matrix
- kernel methods
- support vector
- feature space
- covariance matrix
- kernel space
- principal component analysis
- principal components
- input space
- biologically inspired
- neural fuzzy
- component analysis
- neural model
- hebbian learning
- positive definite
- eigenvalues and eigenvectors
- data sets
- laplacian matrix
- kernel density estimation
- machine learning
- covariance matrices
- learning rules
- feature extraction
- singular value decomposition