A Fast, Well-Founded Approximation to the Empirical Neural Tangent Kernel.
Mohamad Amin MohamadiDanica J. SutherlandPublished in: CoRR (2022)
Keyphrases
- network architecture
- polynomial kernels
- series expansion
- kernel methods
- machine learning
- approximation error
- approximation algorithms
- kernel function
- closed form
- kernel regression
- theoretical analysis
- data sets
- component analysis
- relative error
- gram matrix
- approximation methods
- positive definite
- learning rules
- reproducing kernel hilbert space
- queueing networks
- error bounds
- information theoretic
- support vector
- neural network