Revisiting differentially private linear regression: optimal and adaptive prediction & estimation in unbounded domain
read the original abstract
We revisit the problem of linear regression under a differential privacy constraint. By consolidating existing pieces in the literature, we clarify the correct dependence of the feature, label and coefficient domains in the optimization error and estimation error, hence revealing the delicate price of differential privacy in statistical estimation and statistical learning. Moreover, we propose simple modifications of two existing DP algorithms: (a) posterior sampling, (b) sufficient statistics perturbation, and show that they can be upgraded into **adaptive** algorithms that are able to exploit data-dependent quantities and behave nearly optimally **for every instance**. Extensive experiments are conducted on both simulated data and real data, which conclude that both AdaOPS and AdaSSP outperform the existing techniques on nearly all 36 data sets that we test on.
This paper has not been read by Pith yet.
Forward citations
Cited by 3 Pith papers
-
Differentially Private Nonparametric Confidence Intervals Under Minimal Distributional Assumptions
A black-box resampling procedure produces asymptotically valid and tight differentially private nonparametric confidence intervals for arbitrary quantities from any suitable private estimator.
-
On the Sample Complexity of Differentially Private Policy Optimization
Differential privacy in policy optimization adds sample complexity costs that often appear as lower-order terms rather than dominating the bounds.
-
High-Dimensional Private Linear Regression with Optimal Rates
DP-GD achieves minimax optimal non-asymptotic risk O(γ + γ²/ρ²) for well-conditioned high-dimensional data and power-law scaling for ill-conditioned power-law spectra, with the exponent depending on the privacy parameter ρ.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.