Singular value perturbation and deep network optimization

DOI10.1007/S00365-022-09601-5MaRDI QIDQ2700883zbMATH OpenOpenAlexFDO

Authors Rudolf H. Riedi, Randall Balestriero, Richard G. Baraniuk

Publication date 27 April 2023

Published in Constructive Approximation (Search for Journal in Brave)

Copyright license Creative Commons Attribution 4.0 International

Full work available at URL https://arxiv.org/abs/2203.03099

zbMATH Keywords

deep learningoptimization landscapeperturbation theorysingular values

Mathematics Subject Classification ID

Perturbation theory of linear operators (47A55) Artificial neural networks and deep learning (68T07)

Recommendations

• Convergence analysis of deep residual networks
• Forward stability of ResNet and its variants
• Theoretical issues in deep networks
• Exact solutions of a deep linear network
• Effect of depth and width on local minima in deep learning

Cites work

• Adaptive estimation of a quadratic functional by model selection.
• Deep learning
• Eigenvalues and Condition Numbers of Random Matrices
• Group invariant scattering
• scientific article; zbMATH DE number 1962931 (Why is no real title available?)
• Level-spacing distributions and the Airy kernel
• The right tail exponent of the Tracy-Widom distribution
• The variation of the spectrum of a normal matrix

Cited in

(2)

• scientific article; zbMATH DE number 6923716 (Why is no real title available?)
• Singular Value Perturbation and Deep Network Optimization

This page was built for publication: Singular value perturbation and deep network optimization

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2700883)