Spectral alignment of stochastic gradient descent for high-dimensional classification tasks

DOI10.1214/25-AAP2187MaRDI QIDQ6927564zbMATH OpenFDO

Authors Gerard Ben Arous, Reza Gheissari, Jiaoyang Huang, Aukosh Jagannath

Publication date 18 September 2025

Published in The Annals of Applied Probability (Search for Journal in Brave)

zbMATH Keywords

Gaussian mixture modelmultilayer networkstochastic gradient descentXOR problemspectral alignment

Mathematics Subject Classification ID

Classification and discrimination; cluster analysis (statistical aspects) (62H30) Nonconvex programming, global optimization (90C26)

Cites work

• scientific article; zbMATH DE number 7306931 (Why is no real title available?)
• scientific article; zbMATH DE number 3314813 (Why is no real title available?)
• A Stochastic Approximation Method
• Complexity of random smooth functions on the high-dimensional sphere
• High-dimensional probability. An introduction with applications in data science
• Phase diagram of stochastic gradient descent in high-dimensional two-layer neural networks
• Phase transition of the largest eigenvalue for nonnull complex sample covariance matrices
• Prevalence of neural collapse during the terminal phase of deep learning training
• Random matrices and complexity of spin glasses
• The landscape of empirical risk for nonconvex losses
• The landscape of the spiked tensor model
• The largest eigenvalue of small rank perturbations of Hermitian random matrices

This page was built for publication: Spectral alignment of stochastic gradient descent for high-dimensional classification tasks

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6927564)