English

On Convex Optimization with Semi-Sensitive Features

Machine Learning 2024-06-28 v1 Cryptography and Security Data Structures and Algorithms

Abstract

We study the differentially private (DP) empirical risk minimization (ERM) problem under the semi-sensitive DP setting where only some features are sensitive. This generalizes the Label DP setting where only the label is sensitive. We give improved upper and lower bounds on the excess risk for DP-ERM. In particular, we show that the error only scales polylogarithmically in terms of the sensitive domain size, improving upon previous results that scale polynomially in the sensitive domain size (Ghazi et al., 2021).

Keywords

Cite

@article{arxiv.2406.19040,
  title  = {On Convex Optimization with Semi-Sensitive Features},
  author = {Badih Ghazi and Pritish Kamath and Ravi Kumar and Pasin Manurangsi and Raghu Meka and Chiyuan Zhang},
  journal= {arXiv preprint arXiv:2406.19040},
  year   = {2024}
}

Comments

To appear in COLT 2024

R2 v1 2026-06-28T17:21:02.910Z