Randomized matrix sketching for neural network training and gradient monitoring

DOI10.1142/S0218202526420017MaRDI QIDQ7300358zbMATH OpenFDO

Authors Harbir Antil, Deepanshu Verma

Publication date 20 July 2026

Published in M\(^3\)AS. Mathematical Models \& Methods in Applied Sciences (Search for Journal in Brave)

Full work available at URL https://doi.org/10.1142/S0218202526420017

Mathematics Subject Classification ID

Numerical methods for low-rank matrix approximation; matrix compression (65F55) Numerical optimization and variational techniques (65K10) Artificial neural networks and deep learning (68T07)

Cites work

• Acceleration of Stochastic Approximation by Averaging
• Approximate nearest neighbors and the fast Johnson-Lindenstrauss transform
• Learning representations by back-propagating errors
• Physics-informed neural networks: a deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations
• Practical sketching algorithms for low-rank matrix approximation
• Randomized sketching algorithms for low-memory dynamic optimization
• Streaming low-rank matrix approximation with an application to scientific simulation

Cited in

(1)

• New trends in mathematics for scientific machine learning

This page was built for publication: Randomized matrix sketching for neural network training and gradient monitoring

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7300358)