Spectral bias and task-model alignment explain generalization in kernel regression and infinitely wide neural networks

Canatar et al. propose a predictive theory of generalization in kernel regression applicable to real data. This theory explains various generalization phenomena observed in wide neural networks, which admit a kernel limit and generalize well despite being overparameterized.

Guardado en:

Detalles Bibliográficos
Autores principales:	Abdulkadir Canatar, Blake Bordelon, Cengiz Pehlevan
Formato:	article
Lenguaje:	EN
Publicado:	Nature Portfolio 2021
Materias:	Science Q
Acceso en línea:	https://doaj.org/article/3fb570c6ce05419290b8cc1eebe16977
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:3fb570c6ce05419290b8cc1eebe16977
record_format	dspace
spelling	oai:doaj.org-article:3fb570c6ce05419290b8cc1eebe169772021-12-02T15:52:52ZSpectral bias and task-model alignment explain generalization in kernel regression and infinitely wide neural networks10.1038/s41467-021-23103-12041-1723https://doaj.org/article/3fb570c6ce05419290b8cc1eebe169772021-05-01T00:00:00Zhttps://doi.org/10.1038/s41467-021-23103-1https://doaj.org/toc/2041-1723Canatar et al. propose a predictive theory of generalization in kernel regression applicable to real data. This theory explains various generalization phenomena observed in wide neural networks, which admit a kernel limit and generalize well despite being overparameterized.Abdulkadir CanatarBlake BordelonCengiz PehlevanNature PortfolioarticleScienceQENNature Communications, Vol 12, Iss 1, Pp 1-12 (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	Science Q
spellingShingle	Science Q Abdulkadir Canatar Blake Bordelon Cengiz Pehlevan Spectral bias and task-model alignment explain generalization in kernel regression and infinitely wide neural networks
description	Canatar et al. propose a predictive theory of generalization in kernel regression applicable to real data. This theory explains various generalization phenomena observed in wide neural networks, which admit a kernel limit and generalize well despite being overparameterized.
format	article
author	Abdulkadir Canatar Blake Bordelon Cengiz Pehlevan
author_facet	Abdulkadir Canatar Blake Bordelon Cengiz Pehlevan
author_sort	Abdulkadir Canatar
title	Spectral bias and task-model alignment explain generalization in kernel regression and infinitely wide neural networks
title_short	Spectral bias and task-model alignment explain generalization in kernel regression and infinitely wide neural networks
title_full	Spectral bias and task-model alignment explain generalization in kernel regression and infinitely wide neural networks
title_fullStr	Spectral bias and task-model alignment explain generalization in kernel regression and infinitely wide neural networks
title_full_unstemmed	Spectral bias and task-model alignment explain generalization in kernel regression and infinitely wide neural networks
title_sort	spectral bias and task-model alignment explain generalization in kernel regression and infinitely wide neural networks
publisher	Nature Portfolio
publishDate	2021
url	https://doaj.org/article/3fb570c6ce05419290b8cc1eebe16977
work_keys_str_mv	AT abdulkadircanatar spectralbiasandtaskmodelalignmentexplaingeneralizationinkernelregressionandinfinitelywideneuralnetworks AT blakebordelon spectralbiasandtaskmodelalignmentexplaingeneralizationinkernelregressionandinfinitelywideneuralnetworks AT cengizpehlevan spectralbiasandtaskmodelalignmentexplaingeneralizationinkernelregressionandinfinitelywideneuralnetworks
_version_	1718385565313794048

Spectral bias and task-model alignment explain generalization in kernel regression and infinitely wide neural networks

Ejemplares similares