18
0

Efficient Estimation of a Gaussian Mean with Local Differential Privacy

Abstract

In this paper we study the problem of estimating the unknown mean θ\theta of a unit variance Gaussian distribution in a locally differentially private (LDP) way. In the high-privacy regime (ϵ0.67\epsilon\le 0.67), we identify the exact optimal privacy mechanism that minimizes the variance of the estimator asymptotically. It turns out to be the extraordinarily simple sign mechanism that applies randomized response to the sign of XiθX_i-\theta. However, since this optimal mechanism depends on the unknown mean θ\theta, we employ a two-stage LDP parameter estimation procedure which requires splitting agents into two groups. The first n1n_1 observations are used to consistently but not necessarily efficiently estimate the parameter θ\theta by θ~n1\tilde{\theta}_{n_1}. Then this estimate is updated by applying the sign mechanism with θ~n1\tilde{\theta}_{n_1} instead of θ\theta to the remaining nn1n-n_1 observations, to obtain an LDP and efficient estimator of the unknown mean.

View on arXiv
Comments on this paper