Paper 2026/2075
Bounding the Excess Risk for Linear Models Trained on Marginal-Preserving, Differentially-Private, Synthetic Data
Abstract
The growing use of machine learning (ML) has raised concerns that an ML model may reveal private information about an individual who has contributed to the training dataset. To prevent leakage of sensitive data, we consider using differentially-private (DP), synthetic training data instead of real training data to train an ML model. A key desirable property of synthetic data is its ability to preserve the low-order marginals of the original distribution. Our main contribution comprises novel upper and lower bounds on the excess empirical risk of linear models trained on such synthetic data, for continuous and Lipschitz loss functions. We perform extensive experimentation alongside our theoretical results
Metadata
- Available format(s)
-
PDF
- Category
- Applications
- Publication info
- Published elsewhere. ICML2024
- Keywords
- Differential PrivacyPrivacy-preserving Machine LearningSynthetic Data
- Contact author(s)
- skyzhou @ umd edu
- History
- 2026-09-19: approved
- 2026-09-17: received
- See all versions
- Short URL
- https://ia.cr/2026/2075
- License
-
CC BY-NC
BibTeX
@misc{cryptoeprint:2026/2075,
author = {Yvonne Zhou and Mingyu Liang and Ivan Brugere and Danial Dervovic and Antigoni Polychroniadou and Min Wu and Dana Dachman-Soled},
title = {Bounding the Excess Risk for Linear Models Trained on Marginal-Preserving, Differentially-Private, Synthetic Data},
howpublished = {Cryptology {ePrint} Archive, Paper 2026/2075},
year = {2026},
url = {https://eprint.iacr.org/2026/2075}
}
