Preconditioning meets biased compression for efficient distributed optimization

DOI10.1007/S10287-023-00496-6MaRDI QIDQ6149587zbMATH OpenOpenAlexFDO

Authors Vitali Pirau, Aleksandr Beznosikov, M. Takáč, A. V. Gasnikov

Publication date 6 February 2024

Published in Computational Management Science (Search for Journal in Brave)

Full work available at URL https://doi.org/10.1007/s10287-023-00496-6

zbMATH Keywords

first-order methodsnon-convex minimizationaccelerated methods

Mathematics Subject Classification ID

Operations research and management science (90Bxx)

Recommendations

• Stochastic gradient methods with preconditioned updates
• Hyperfast second-order local solvers for efficient statistically preconditioned distributed optimization
• Stochastic distributed learning with gradient quantization and double-variance reduction
• Communication-efficient distributed optimization of self-concordant empirical loss
• On data preconditioning for regularized loss minimization

Cites work

• Adaptive subgradient methods for online learning and stochastic optimization
• An estimator for the diagonal of a matrix
• Lectures on convex optimization
• Numerical Optimization
• Understanding machine learning. From theory to algorithms

Cited in

(3)

• Editorial
• Hyperfast second-order local solvers for efficient statistically preconditioned distributed optimization
• Stochastic gradient methods with preconditioned updates

This page was built for publication: Preconditioning meets biased compression for efficient distributed optimization

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6149587)