On the privacy of noisy stochastic gradient descent for convex optimization

DOI10.1137/23M1556538MaRDI QIDQ6583673zbMATH OpenFDO

Authors Jason M. Altschuler, Jinho Bok, Kunal Talwar

Publication date 6 August 2024

Published in SIAM Journal on Computing (Search for Journal in Brave)

zbMATH Keywords

convex optimizationstochastic gradient Langevin dynamicsdifferential privacynoisy stochastic gradient descentprivacy amplification by iterationshifted divergences

Mathematics Subject Classification ID

Convex programming (90C25) Probability in computer science (algorithm analysis, random structures, phase transitions, etc.) (68Q87) Privacy of data (68P27)

Cites work

• Convex optimization: algorithms and complexity
• Differentially private empirical risk minimization
• High-dimensional Bayesian inference via the unadjusted Langevin algorithm
• Private selection from private candidates
• Private stochastic convex optimization: optimal rates in linear time
• Probability and computing. Randomization and probabilistic techniques in algorithms and data analysis
• Probability with Martingales
• Rényi Divergence and Kullback-Leibler Divergence
• Sampling from a log-concave distribution with projected Langevin Monte Carlo
• Theoretical Guarantees for Approximate Sampling from Smooth and Log-Concave Densities
• Theory of Cryptography
• What can we learn privately?

This page was built for publication: On the privacy of noisy stochastic gradient descent for convex optimization

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6583673)