mlsanityData to 14 Oct 2026

Enhancing Differentially Private Mechanisms via Empirical Bayes

arXiv 2606.21107 ↗First posted 19 Jun 2026cs.LG cs.CR stat.MEcited by 0

Not on the map. Why?

Abstract (arXiv)

Differential privacy (DP) has become the gold standard for ensuring the privacy protection of machine learning and statistical algorithms in recent decades. A plethora of algorithms and methods have been developed to enhance the utility of DP algorithms while maintaining the same level of DP. However, these are often overly complex or computationally ineffective. We propose a novel approach focusing on denoising the output of the simple additive Gaussian mechanism by adopting the idea of textitempirical Bayes estimation. We highlight that the empirical Bayes approach can reduce the mean-squared error solely by taking the output of the Gaussian mechanism as input. Our numerical studies show that this simple yet powerful approach can be applied to improve upon various statistical problems, including histogram release, principal component analysis, and linear regression, often outperforming existing private algorithms.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 20 of its 46 references that are in the map. First posted on arXiv 19 Jun 2026.

Builds on

20 of 46 references.

References · 20

  1. 2016Deep Learning with Differential Privacycited by 2,728
  2. 2016Concentrated Differential Privacy: Simplifications, Extensions, and Lower Boundscited by 346
  3. 2018Improving the Gaussian Mechanism for Differential Privacy: Analytical Calibration and Optimal Denoisingcited by 218
  4. 2020Understanding Gradient Clipping in Private SGD: A Geometric Perspectivecited by 142
  5. 2018Revisiting differentially private linear regression: optimal and adaptive prediction & estimation in unbounded domaincited by 61
  6. 2013The Noisy Power Method: A Meta Algorithm with Applicationscited by 58
  7. 2016Rate-Optimal Perturbation Bounds for Singular Subspaces with Applications to High-Dimensional Statisticscited by 45
  8. 2020Improving Deep Learning with Differential Privacy using Gradient Encoding and Denoisingcited by 30
  9. 2023From Robustness to Privacy and Backcited by 27
  10. 2022DP-PCA: Statistically Optimal and Differentially Private PCAcited by 24
  11. 2022Easy Differentially Private Linear Regressioncited by 19
  12. 2024Private Gradient Descent for Linear Regression: Tighter Error Bounds and Instance-Specific Uncertainty Estimationcited by 12
8 more
  1. 2024Optimal Differentially Private PCA and Estimation for Spiked Covariance Matricescited by 11
  2. 2021DPlis: Boosting Utility of Differentially Private Deep Learning via Randomized Smoothingcited by 10
  3. 2014Lognormal Distributions and Geometric Averages of Positive Definite Matricescited by 7
  4. 2017Differentially Private Learning of Undirected Graphical Models using Collective Graphical Modelscited by 6
  5. 2023Improved Differentially Private Regression via Gradient Boostingcited by 6
  6. 2022Stochastic and Private Nonconvex Outlier-Robust PCAcited by 4
  7. 2022Differentially Private Multivariate Statistics with an Application to Contingency Table Analysiscited by 2
  8. 2023On the Utility Gain of Iterative Bayesian Update for Locally Differentially Private Mechanismscited by 1

Built on by

Not cited yet.