Generalized PTR: User-Friendly Recipes for Data-Adaptive Algorithms with Differential Privacy
Rachel Redberg, Yuqing Zhu, Yu-Xiang Wang
Abstract
The “Propose-Test-Release” (PTR) framework [Dwork and Lei, 2009] is a classic recipe for designing differentially private (DP) algorithms that are data-adaptive, i.e. those that add less noise when the input dataset is “nice”. We extend PTR to a more general setting by privately testing data-dependent privacy losses rather than local sensitivity, hence making it applicable beyond the standard noise-adding mechanisms, e.g. to queries with unbounded or undefined sensitivity. We demonstrate the versatility of generalized PTR using private linear regression as a case study. Additionally, we apply our algorithm to solve an open problem from “Private Aggregation of Teacher Ensembles (PATE)” [Papernot et al., 2017, 2018] - privately releasing the entire model with a delicate data-dependent analysis.
BibTeX
@InProceedings{pmlr-v206-redberg23a,
title = {Generalized PTR: User-Friendly Recipes for Data-Adaptive Algorithms with Differential Privacy},
author = {Redberg, Rachel and Zhu, Yuqing and Wang, Yu-Xiang},
booktitle = {Proceedings of The 26th International Conference on Artificial Intelligence and Statistics},
pages = {3977--4005},
year = {2023},
editor = {Ruiz, Francisco and Dy, Jennifer and van de Meent, Jan-Willem},
volume = {206},
series = {Proceedings of Machine Learning Research},
month = {25--27 Apr},
publisher = {PMLR},
pdf = {https://proceedings.mlr.press/v206/redberg23a/redberg23a.pdf},
url = {https://proceedings.mlr.press/v206/redberg23a.html},
abstract = {The “Propose-Test-Release” (PTR) framework [Dwork and Lei, 2009] is a classic recipe for designing differentially private (DP) algorithms that are data-adaptive, i.e. those that add less noise when the input dataset is “nice”. We extend PTR to a more general setting by privately testing data-dependent privacy losses rather than local sensitivity, hence making it applicable beyond the standard noise-adding mechanisms, e.g. to queries with unbounded or undefined sensitivity. We demonstrate the versatility of generalized PTR using private linear regression as a case study. Additionally, we apply our algorithm to solve an open problem from “Private Aggregation of Teacher Ensembles (PATE)” [Papernot et al., 2017, 2018] - privately releasing the entire model with a delicate data-dependent analysis.}
}