Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Optimal Algorithms for Mean Estimation under Local Differential Privacy

About

We study the problem of mean estimation of $\ell_2$-bounded vectors under the constraint of local differential privacy. While the literature has a variety of algorithms that achieve the asymptotically optimal rates for this problem, the performance of these algorithms in practice can vary significantly due to varying (and often large) hidden constants. In this work, we investigate the question of designing the protocol with the smallest variance. We show that PrivUnit (Bhowmick et al. 2018) with optimized parameters achieves the optimal variance among a large family of locally private randomizers. To prove this result, we establish some properties of local randomizers, and use symmetrization arguments that allow us to write the optimal randomizer as the optimizer of a certain linear program. These structural results, which should extend to other problems, then allow us to show that the optimal randomizer belongs to the PrivUnit family. We also develop a new variant of PrivUnit based on the Gaussian distribution which is more amenable to mathematical analysis and enjoys the same optimality guarantees. This allows us to establish several useful properties on the exact constants of the optimal error as well as to numerically estimate these constants.

Hilal Asi, Vitaly Feldman, Kunal Talwar• 2022

Related benchmarks

TaskDatasetResultRank
Linear ClassificationCIFAR-10 (test)
Top-1 Accuracy53.92
116
Linear regressionLHSM (test)
RMSE0.3365
108
Linear regressionLHSM m=16 (test)
RMSE0.3365
108
Image ClassificationCIFAR-10-C Brightness Corruption, Severity Level 5 (test)
Accuracy (CIFAR-10-C Brightness 5)33.75
107
Image ClassificationMNIST VAE m=32 latent space (test)
Accuracy46.82
85
Image ClassificationCIFAR-10
Accuracy (ε=0.5)10.44
10
Showing 6 of 6 rows

Other info

Follow for update