Neural tangent kernel at initialization: linear width suffices.
Arindam BanerjeePedro Cisneros-VelardeLibin ZhuMikhail BelkinPublished in: UAI (2023)
Keyphrases
- kernel function
- neural network
- support vector
- network architecture
- discriminant function
- k means
- reproducing kernel hilbert space
- mutual subspace method
- regularized least squares
- multiscale
- kernel methods
- gaussian processes
- euclidean space
- hilbert space
- simple linear
- neural model
- multiple kernel learning
- bio inspired
- input space
- support vector machine
- feature space
- similarity measure
- genetic algorithm