Characterizing dynamical stability of stochastic gradient descent in overparameterized learning

MaRDI QIDQ6887426zbMATH OpenFDO

Authors Dennis Chemnitz, Maximilian Engel

Publication date 9 December 2025

Published in Journal of Machine Learning Research (JMLR) (Search for Journal in Brave)

Copyright license Creative Commons Attribution 4.0 International

Full work available at URL https://jmlr.org/papers/v26/24-1547.html

zbMATH Keywords

linear stabilityLyapunov exponentsmachine learningoverparameterizationstochastic gradient descent

Mathematics Subject Classification ID

Learning and adaptive systems in artificial intelligence (68T05) Stochastic programming (90C15)

Cites work

• A Formula Connecting Sample and Moment Stability of Linear Stochastic Systems
• A noise-induced transition in the Lorenz system
• Almost-sure exponential mixing of passive scalars by the stochastic Navier-Stokes equations
• Convergence rates for the stochastic gradient descent method for non-convex objective functions
• Ergodic theory of chaos and strange attractors
• Exponential mixing for random dynamical systems and an example of Pierrehumbert
• Flat Minima
• Global minima of overparameterized neural networks
• scientific article; zbMATH DE number 3913325 (Why is no real title available?)
• scientific article; zbMATH DE number 3944973 (Why is no real title available?)
• scientific article; zbMATH DE number 4028507 (Why is no real title available?)
• scientific article; zbMATH DE number 3709275 (Why is no real title available?)
• scientific article; zbMATH DE number 19400 (Why is no real title available?)
• scientific article; zbMATH DE number 1201579 (Why is no real title available?)
• scientific article; zbMATH DE number 1950480 (Why is no real title available?)
• scientific article; zbMATH DE number 3798745 (Why is no real title available?)
• scientific article; zbMATH DE number 776267 (Why is no real title available?)
• scientific article; zbMATH DE number 3290003 (Why is no real title available?)
• Hypoelliptic second order differential equations
• Kingman's subadditive ergodic theorem
• Large deviations and stochastic flows of diffeomorphisms
• Learning representations by back-propagating errors
• Markov chains on metric spaces. A short course
• Multilayer feedforward networks are universal approximators
• Perturbation theory for linear operators.
• Smooth ergodic theory for endomorphisms
• Smooth ergodic theory of random dynamical systems
• Stochastic modified equations and dynamics of stochastic gradient algorithms. I: Mathematical foundations

This page was built for publication: Characterizing dynamical stability of stochastic gradient descent in overparameterized learning

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6887426)