Cargando…
Spectral bias and task-model alignment explain generalization in kernel regression and infinitely wide neural networks
A theoretical understanding of generalization remains an open problem for many machine learning models, including deep networks where overparameterization leads to better performance, contradicting the conventional wisdom from classical statistics. Here, we investigate generalization error for kerne...
Autores principales: | Canatar, Abdulkadir, Bordelon, Blake, Pehlevan, Cengiz |
---|---|
Formato: | Online Artículo Texto |
Lenguaje: | English |
Publicado: |
Nature Publishing Group UK
2021
|
Materias: | |
Acceso en línea: | https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8131612/ https://www.ncbi.nlm.nih.gov/pubmed/34006842 http://dx.doi.org/10.1038/s41467-021-23103-1 |
Ejemplares similares
-
Population codes enable learning from few examples by shaping inductive bias
por: Bordelon, Blake, et al.
Publicado: (2022) -
Spectral methods in infinite-dimensional analysis
por: Berezansky, Y M, et al.
Publicado: (1995) -
Spectral theory of infinite-area hyperbolic surfaces
por: Borthwick, David
Publicado: (2016) -
Spectral theory of infinite-area hyperbolic surfaces
por: Borthwick, David
Publicado: (2007) -
Limit operators, collective compactness, and the spectral theory of infinite matrices
por: Chandler-Wilde, Simon N, et al.
Publicado: (2011)