Matches in SemOpenAlex for { <https://semopenalex.org/work/W3034708079> ?p ?o ?g. }
Showing items 1 to 92 of
92
with 100 items per page.
- W3034708079 endingPage "1034" @default.
- W3034708079 startingPage "1024" @default.
- W3034708079 abstract "We derive analytical expressions for the generalization performance of kernel regression as a function of the number of training samples using theoretical methods from Gaussian processes and statistical physics. Our expressions apply to wide neural networks due to an equivalence between training them and kernel regression with the Neural Tangent Kernel (NTK). By computing the decomposition of the total generalization error due to different spectral components of the kernel, we identify a new spectral principle: as the size of the training set grows, kernel machines and neural networks fit successively higher spectral modes of the target function. When data are sampled from a uniform distribution on a high-dimensional hypersphere, dot product kernels, including NTK, exhibit learning stages where different frequency modes of the target function are learned. We verify our theory with simulations on synthetic data and MNIST dataset." @default.
- W3034708079 created "2020-06-19" @default.
- W3034708079 creator A5023195984 @default.
- W3034708079 creator A5039282308 @default.
- W3034708079 creator A5040593057 @default.
- W3034708079 date "2020-07-12" @default.
- W3034708079 modified "2023-09-23" @default.
- W3034708079 title "Spectrum Dependent Learning Curves in Kernel Regression and Wide Neural Networks" @default.
- W3034708079 hasPublicationYear "2020" @default.
- W3034708079 type Work @default.
- W3034708079 sameAs 3034708079 @default.
- W3034708079 citedByCount "16" @default.
- W3034708079 countsByYear W30347080792020 @default.
- W3034708079 countsByYear W30347080792021 @default.
- W3034708079 crossrefType "proceedings-article" @default.
- W3034708079 hasAuthorship W3034708079A5023195984 @default.
- W3034708079 hasAuthorship W3034708079A5039282308 @default.
- W3034708079 hasAuthorship W3034708079A5040593057 @default.
- W3034708079 hasConcept C114614502 @default.
- W3034708079 hasConcept C121332964 @default.
- W3034708079 hasConcept C122280245 @default.
- W3034708079 hasConcept C12267149 @default.
- W3034708079 hasConcept C134306372 @default.
- W3034708079 hasConcept C134517425 @default.
- W3034708079 hasConcept C153180895 @default.
- W3034708079 hasConcept C154945302 @default.
- W3034708079 hasConcept C163716315 @default.
- W3034708079 hasConcept C177148314 @default.
- W3034708079 hasConcept C182335926 @default.
- W3034708079 hasConcept C190502265 @default.
- W3034708079 hasConcept C195699287 @default.
- W3034708079 hasConcept C27438332 @default.
- W3034708079 hasConcept C2776562905 @default.
- W3034708079 hasConcept C33923547 @default.
- W3034708079 hasConcept C41008148 @default.
- W3034708079 hasConcept C50644808 @default.
- W3034708079 hasConcept C61326573 @default.
- W3034708079 hasConcept C62520636 @default.
- W3034708079 hasConcept C7218915 @default.
- W3034708079 hasConcept C74193536 @default.
- W3034708079 hasConcept C74887250 @default.
- W3034708079 hasConceptScore W3034708079C114614502 @default.
- W3034708079 hasConceptScore W3034708079C121332964 @default.
- W3034708079 hasConceptScore W3034708079C122280245 @default.
- W3034708079 hasConceptScore W3034708079C12267149 @default.
- W3034708079 hasConceptScore W3034708079C134306372 @default.
- W3034708079 hasConceptScore W3034708079C134517425 @default.
- W3034708079 hasConceptScore W3034708079C153180895 @default.
- W3034708079 hasConceptScore W3034708079C154945302 @default.
- W3034708079 hasConceptScore W3034708079C163716315 @default.
- W3034708079 hasConceptScore W3034708079C177148314 @default.
- W3034708079 hasConceptScore W3034708079C182335926 @default.
- W3034708079 hasConceptScore W3034708079C190502265 @default.
- W3034708079 hasConceptScore W3034708079C195699287 @default.
- W3034708079 hasConceptScore W3034708079C27438332 @default.
- W3034708079 hasConceptScore W3034708079C2776562905 @default.
- W3034708079 hasConceptScore W3034708079C33923547 @default.
- W3034708079 hasConceptScore W3034708079C41008148 @default.
- W3034708079 hasConceptScore W3034708079C50644808 @default.
- W3034708079 hasConceptScore W3034708079C61326573 @default.
- W3034708079 hasConceptScore W3034708079C62520636 @default.
- W3034708079 hasConceptScore W3034708079C7218915 @default.
- W3034708079 hasConceptScore W3034708079C74193536 @default.
- W3034708079 hasConceptScore W3034708079C74887250 @default.
- W3034708079 hasOpenAccess W3034708079 @default.
- W3034708079 hasRelatedWork W1567512734 @default.
- W3034708079 hasRelatedWork W1748633101 @default.
- W3034708079 hasRelatedWork W1957492789 @default.
- W3034708079 hasRelatedWork W2809090039 @default.
- W3034708079 hasRelatedWork W2910655610 @default.
- W3034708079 hasRelatedWork W2922026128 @default.
- W3034708079 hasRelatedWork W2923764619 @default.
- W3034708079 hasRelatedWork W2952204734 @default.
- W3034708079 hasRelatedWork W2963518130 @default.
- W3034708079 hasRelatedWork W2964052793 @default.
- W3034708079 hasRelatedWork W2967536008 @default.
- W3034708079 hasRelatedWork W2970217468 @default.
- W3034708079 hasRelatedWork W2971043187 @default.
- W3034708079 hasRelatedWork W2994747787 @default.
- W3034708079 hasRelatedWork W3005509776 @default.
- W3034708079 hasRelatedWork W3039537761 @default.
- W3034708079 hasRelatedWork W3101069636 @default.
- W3034708079 hasRelatedWork W3117575511 @default.
- W3034708079 hasRelatedWork W3137695714 @default.
- W3034708079 hasRelatedWork W3162003518 @default.
- W3034708079 hasVolume "1" @default.
- W3034708079 isParatext "false" @default.
- W3034708079 isRetracted "false" @default.
- W3034708079 magId "3034708079" @default.
- W3034708079 workType "article" @default.