Communication Complexity in Locally Private Distribution Estimation and Heavy Hitters

[edit]

Jayadev Acharya, Ziteng Sun ;
Proceedings of the 36th International Conference on Machine Learning, PMLR 97:51-60, 2019.

Abstract

We consider the problems of distribution estimation, and heavy hitter (frequency) estimation under privacy, and communication constraints. While the constraints have been studied separately, optimal schemes for one are sub-optimal for the other. We propose a sample-optimal $\eps$-locally differentially private (LDP) scheme for distribution estimation, where each user communicates one bit, and requires no public randomness. We also show that Hadamard Response, a recently proposed scheme for $\eps$-LDP distribution estimation is also utility-optimal for heavy hitters estimation. Our final result shows that unlike distribution estimation, without public randomness, any utility-optimal heavy hitter estimation algorithm must require $\Omega(\log n)$ bits of communication per user.

Related Material