• Corpus ID: 239998433

Unbiased Statistical Estimation and Valid Confidence Intervals Under Differential Privacy

  title={Unbiased Statistical Estimation and Valid Confidence Intervals Under Differential Privacy},
  author={Christian Covington and Xi He and James Honaker and Gautam Kamath},
We present a method for producing unbiased parameter estimates and valid confidence intervals under the constraints of differential privacy, a formal framework for limiting individual information leakage from sensitive data. Prior work in this area is limited in that it is tailored to calculating confidence intervals for specific statistical procedures, such as mean estimation or simple linear regression. While other recent work can produce confidence intervals for more general sets of… 

Figures and Tables from this paper

Privacy Preserving Inference on the Ratio of Two Gaussians Using (Weighted) Sums
The delta method is used to derive the asymptotic distribution of the ratio estimator and the Gaussian mechanism to provide (ǫ, δ) privacy guarantees and it is shown that the CIs of the methods have the right coverage with proper privacy budget.
Differentially private inference via noisy optimization
This work shows that robust statistics can be used in conjunction with noisy gradient descent or noisy Newton methods in order to obtain optimal private estimators with global linear or quadratic convergence, respectively, and establishes local and global convergence guarantees.
Privacy-Preserving Randomized Controlled Trials: A Protocol for Industry Scale Deployment
This paper outlines a way to deploy an end-to-end privacy-preserving protocol for learning causal effects from Randomized Controlled Trials, particularly focused on the difficult and important case where one party determines which treatment an individual receives, and another party measures outcomes on individuals, and these parties do not want to leak any of their information to each other.


Non-parametric Differentially Private Confidence Intervals for the Median
This paper proposes and evaluates several strategies to compute valid differentially private confidence intervals for the median and illustrates that addressing both sources of uncertainty–the error from sampling and the error from protecting the output–simultaneously should be preferred over simpler approaches that incorporate the uncertainty in a sequential fashion.
General-Purpose Differentially-Private Confidence Intervals
This work develops two broadly applicable methods for private confidence-interval construction based on asymptotics and the parametric bootstrap, which applies "out of the box" to a wide class of private estimators and has good coverage at small sample sizes, but with increased computational cost.
Bootstrap Inference and Differential Privacy: Standard Errors for Free∗
The bootstrap is a common and powerful statistical tool for numerically computing the standard error of estimators, that is, a calculation of the uncertainty of functions computed on sample data so
Differentially private model selection with penalized and constrained likelihood
This work shows that model selection procedures based on penalized least squares or likelihood can be made differentially private by a combination of regularization and randomization, and proposes two algorithms to do so.
Differentially Private Inference for Binomial Data
A 'Neyman-Pearson lemma' for binomial data under DP is proved, which proves that in general, DP hypothesis tests can be written in terms of linear constraints, and for exchangeable data can always be expressed as a function of the empirical distribution.
Revisiting Differentially Private Hypothesis Tests for Categorical Data
A modified equivalence between chi-squared tests and likelihood ratio tests is shown, more suited to hypothesis testing with privacy, and differentially private likelihood ratio and chi-Squared tests for a variety of applications on tabular data are developed.
Statistically Valid Inferences from Privacy Protected Data
The vast majority of data that could help social scientists understand and ameliorate the challenges of human society is presently locked away inside companies, in part because of worries about
The structure of optimal private tests for simple hypotheses
Hypothesis testing plays a central role in statistical inference, and is used in many settings where privacy concerns are paramount. This work answers a basic question about privately testing simple
Differentially private methods for managing model uncertainty in linear regression models
In this article, we develop differentially private tools for handling model uncertainty in linear regression models. We introduce hypothesis tests for nested linear models and methods for model
Finite Sample Differentially Private Confidence Intervals
These algorithms guarantee a finite sample coverage, as opposed to an asymptotic coverage, and prove lower bounds on the expected size of any differentially private confidence set showing that the parameters are optimal up to polylogarithmic factors.