Select without fear: almost all minibatch schedules generalize optimally

DOI10.1137/23M1617096MaRDI QIDQ6944076zbMATH OpenFDO

Authors Konstantinos E. Nikolakakis, Amin Karbasi

Publication date 12 August 2025

Published in SIAM Journal on Mathematics of Data Science (Search for Journal in Brave)

Full work available at URL https://doi.org/10.1137/23M1617096

zbMATH Keywords

convex optimizationgeneralization errorminimax boundssmooth nonconvex optimization

Mathematics Subject Classification ID

Learning and adaptive systems in artificial intelligence (68T05) Stochastic programming (90C15) Convex programming (90C25) Nonconvex programming, global optimization (90C26)

Cites work

• 10.1162/153244302760200704
• Black-box tests for algorithmic stability
• Convergence rate of incremental gradient and incremental Newton methods
• Convergence rate of incremental subgradient algorithms
• Convex optimization algorithms
• Generalization performance of multi-pass stochastic gradient descent with convex loss functions
• scientific article; zbMATH DE number 1818892 (Why is no real title available?)
• Limitations of information-theoretic generalization bounds for gradient descent methods in stochastic convex optimization
• On the Convergence Rate of Incremental Aggregated Gradient Algorithms
• Optimization and learning with randomly compressed gradient updates
• Stochastic nonconvex optimization with large minibatches
• Understanding generalization error of SGD in nonconvex optimization
• Why random reshuffling beats stochastic gradient descent

This page was built for publication: Select without fear: almost all minibatch schedules generalize optimally

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6944076)