mlsanityData to 14 Oct 2026

Noise-Calibrated Inference from Differentially Private Sufficient Statistics in Exponential Families

arXiv 2603.02010 ↗First posted 2 Mar 2026cs.LG stat.MLcited by 0

Not on the map. Why?

Abstract (arXiv)

Many differentially private (DP) data release systems either output DP synthetic data and leave analysts to perform inference as usual, which can lead to severe miscalibration, or output a DP point estimate without a principled way to do uncertainty quantification. This paper develops a clean and tractable middle ground for exponential families: release only DP sufficient statistics, then perform noise-calibrated likelihood-based inference and optional parametric synthetic data generation as post-processing. Our contributions are: (1) a general recipe for approximate-DP release of clipped sufficient statistics under the Gaussian mechanism; (2) asymptotic normality, explicit variance inflation, and valid Wald-style confidence intervals for the plug-in DP MLE; (3) a noise-aware likelihood correction that is first-order equivalent to the plug-in but supports bootstrap-based intervals; and (4) a matching minimax lower bound showing the privacy distortion rate is unavoidable. The resulting theory yields concrete design rules and a practical pipeline for releasing DP synthetic data with principled uncertainty quantification, validated on three exponential families and real census data.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 18 of its 28 references that are in the map. First posted on arXiv 2 Mar 2026.

Builds on

18 of 28 references.

References · 18

  1. 2021Retiring Adult: New Datasets for Fair Machine Learningcited by 449
  2. 2018Improving the Gaussian Mechanism for Differential Privacy: Analytical Calibration and Optimal Denoisingcited by 218
  3. 2016Minimax Optimal Procedures for Locally Private Estimationcited by 107
  4. 2017Finite Sample Differentially Private Confidence Intervalscited by 104
  5. 2021Winning the NIST Contest: A scalable and general approach to differentially private synthetic datacited by 102
  6. 2022AIM: An Adaptive and Iterative Mechanism for Differentially Private Synthetic Datacited by 90
  7. 2021Benchmarking Differentially Private Synthetic Data Generation Algorithmscited by 81
  8. 2021Differentially private inference via noisy optimizationcited by 27
  9. 2020Parametric Bootstrap for Differentially Private Confidence Intervalscited by 19
  10. 2018Differentially Private Bayesian Inference for Exponential Familiescited by 18
  11. 2021Unbiased Statistical Estimation and Valid Confidence Intervals Under Differential Privacycited by 15
  12. 2023A Bias-Accuracy-Privacy Trilemma for Statistical Estimationcited by 12
6 more
  1. 2022Differentially Private Bootstrap: New Privacy Analysis and Inference Strategiescited by 10
  2. 2022Noise-Aware Statistical Inference with Differentially Private Synthetic Datacited by 9
  3. 2024Privacy-preserving data release leveraging optimal transport and particle gradient descentcited by 8
  4. 2022Hypothesis Testing for Differentially Private Linear Regressioncited by 7
  5. 2023Simulation-based, Finite-sample Inference for Privatized Datacited by 7
  6. 2024Does Differentially Private Synthetic Data Lead to Synthetic Discoveries?cited by 4

Built on by

Not cited yet.