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.
Cite
Text
Zhou et al. "Bounding the Excess Risk for Linear Models Trained on Marginal-Preserving, Differentially-Private, Synthetic Data." International Conference on Machine Learning, 2024.Markdown
[Zhou et al. "Bounding the Excess Risk for Linear Models Trained on Marginal-Preserving, Differentially-Private, Synthetic Data." International Conference on Machine Learning, 2024.](https://mlanthology.org/icml/2024/zhou2024icml-bounding/)BibTeX
@inproceedings{zhou2024icml-bounding,
title = {{Bounding the Excess Risk for Linear Models Trained on Marginal-Preserving, Differentially-Private, Synthetic Data}},
author = {Zhou, Yvonne and Liang, Mingyu and Brugere, Ivan and Dervovic, Danial and Polychroniadou, Antigoni and Wu, Min and Dachman-Soled, Dana},
booktitle = {International Conference on Machine Learning},
year = {2024},
pages = {61979-62001},
volume = {235},
url = {https://mlanthology.org/icml/2024/zhou2024icml-bounding/}
}