Neural Tangent Kernel

GPTKB entity

Statements (33)
Predicate Object
gptkbp:instanceOf gptkb:mathematical_concept
learning theory
gptkbp:abbreviation gptkb:NTK
gptkbp:appliesTo convolutional neural networks
recurrent neural networks
fully connected neural networks
gptkbp:arXivID 1806.07572
gptkbp:basisFor NTK regression
NTK-based generalization bounds
NTK-based optimization analysis
gptkbp:describes infinite-width neural networks
gptkbp:enables closed-form solutions for training dynamics
linearization of neural network training
gptkbp:field gptkb:machine_learning
gptkb:mathematics
deep learning
https://www.w3.org/2000/01/rdf-schema#label Neural Tangent Kernel
gptkbp:influenced theoretical deep learning research
gptkbp:introduced Arthur Jacot
Clément Hongler
Franck Gabriel
gptkbp:introducedIn 2018
gptkbp:limitingCaseOf infinite-width neural networks
gptkbp:publishedIn Neural Tangent Kernel: Convergence and Generalization in Neural Networks
gptkbp:relatedTo gptkb:Gaussian_processes
gptkb:kernel_methods
gradient descent
function space view
random initialization
gptkbp:usedFor theoretical analysis of neural networks
analyzing training dynamics
gptkbp:bfsParent gptkb:NTK
gptkbp:bfsLayer 5