Abstract:Privacy-protecting data analysis investigates statistical methods under privacy constraints. This is a rising challenge in modern statistics, as the achievement of confidentiality guarantees, which typically occurs through suitable perturbations of the data, may determine a loss in the statistical utility of the data. In this paper, we consider privacy-protecting tests for goodness-of-fit in frequency tables, this being arguably the most common form of releasing data, and present a rigorous analysis of the large sample behaviour of a private likelihood-ratio (LR) test. Under the framework of $(\varepsilon,\delta)$-differential privacy for perturbed data, our main contribution is the power analysis of the private LR test, which characterizes the trade-off between confidentiality, measured via the differential privacy parameters $(\varepsilon,\delta)$, and statistical utility, measured via the power of the test. This is obtained through a Bahadur-Rao large deviation expansion for the power of the private LR test, bringing out a critical quantity, as a function of the sample size, the dimension of the table and $(\varepsilon,\delta)$, that determines a loss in the power of the test. Such a result is then applied to characterize the impact of the sample size and the dimension of the table, in connection with the parameters $(\varepsilon,\delta)$, on the loss of the power of the private LR test. In particular, we determine the (sample) cost of $(\varepsilon,\delta)$-differential privacy in the private LR test, namely the additional sample size that is required to recover the power of the Multinomial LR test in the absence of perturbation. Our power analysis rely on a non-standard large deviation analysis for the LR, as well as the development of a novel (sharp) large deviation principle for sum of i.i.d. random vectors, which is of independent interest.

Differentially Private Hypothesis Testing, Revisited.

Revisiting Differentially Private Hypothesis Tests for Categorical Data

Differentially Private Chi-Squared Hypothesis Testing: Goodness of Fit and Independence Testing

Differentially private Bayesian tests

Differentially Private Kolmogorov-Smirnov-Type Tests

Differentially Private Permutation Tests: Applications to Kernel Methods

Hypothesis Testing Interpretations and Renyi Differential Privacy

A Statistical Viewpoint on Differential Privacy: Hypothesis Testing, Representation and Blackwell's Theorem

Private Sequential Hypothesis Testing for Statisticians: Privacy, Error Rates, and Sample Size

Improved Differentially Private Analysis of Variance

Statistical Approximating Distributions under Differential Privacy.

Locally Private Hypothesis Testing

Statistical Inference in the Differential Privacy Model

Private High-Dimensional Hypothesis Testing

Differentially Private Multivariate Statistics with an Application to Contingency Table Analysis

The power of private likelihood-ratio tests for goodness-of-fit in frequency tables

Minimax Optimal Two-Sample Testing under Local Differential Privacy

Federated Nonparametric Hypothesis Testing with Differential Privacy Constraints: Optimal Rates and Adaptive Tests

Differentially Private Algorithms for Empirical Machine Learning

Revisiting Differentially Private Hyper-parameter Tuning

Differential Privacy: An Economic Method for Choosing Epsilon