Old techniques in differentially private linear regression

MaRDI QIDQ7034371zbMATH OpenFDO

Authors Or Sheffet

Publication date 31 January 2025

Full work available at URL https://proceedings.mlr.press/v98/sheffet19a.html

zbMATH Keywords

differential privacylinear regressionsecond-moment matrixWishart distribution

Mathematics Subject Classification ID

Linear regression; mixed models (62J05) Ridge regression; shrinkage estimators (Lasso) (62J07) Statistical aspects of big data and data science (62R07) Privacy of data (68P27)

Cites work

• A risk comparison of ordinary least squares vs ridge regression
• An elementary proof of a theorem of Johnson and Lindenstrauss
• Analyze Gauss: optimal bounds for privacy-preserving principal component analysis
• Beating randomized response on incoherent matrices
• Beyond worst-case analysis in private singular vector computation
• Boosting and differential privacy
• Concentrated differential privacy: simplifications, extensions, and lower bounds
• Differentially private empirical risk minimization
• scientific article; zbMATH DE number 3673370 (Why is no real title available?)
• scientific article; zbMATH DE number 6026126 (Why is no real title available?)
• scientific article; zbMATH DE number 3227378 (Why is no real title available?)
• Local operator theory, random matrices and Banach spaces.
• On differentially private low rank approximation
• Our Data, Ourselves: Privacy Via Distributed Noise Generation
• Private empirical risk minimization: efficient algorithms and tight error bounds
• Ridge Regression: Biased Estimation for Nonorthogonal Problems
• Robust and private Bayesian inference
• The Elements of Statistical Learning
• The Johnson-Lindenstrauss transform itself preserves differential privacy
• Theory of Cryptography

Cited in

(3)

• Private low-rank approximation for covariance matrices, Dyson Brownian motion, and eigenvalue-gap bounds for Gaussian perturbations
• Differentially private learning beyond the classical dimensionality regime
• Privacy guarantees in posterior sampling under contamination

This page was built for publication: Old techniques in differentially private linear regression

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7034371)