Locally Private $k$-Means Clustering with Constant Multiplicative Approximation and Near-Optimal Additive Error

TL;DR

This paper introduces two new algorithms for locally private $k$-means clustering that achieve near-optimal additive error bounds and constant multiplicative approximation, improving upon previous bounds and demonstrating polynomial dependence on $k$.

Contribution

The authors develop two algorithms that significantly reduce the additive error in locally private $k$-means clustering, achieving near-linear bounds and constant multiplicative approximation with a constant number of rounds.

Findings

First algorithm achieves $(1+ ext{small})$-approximation with $ ilde{O}( ext{poly}(1/ ext{alpha}) imes ext{sqrt}(d'n))$ additive error.

Second algorithm achieves $O( ext{poly}(k) imes ext{sqrt}(d'n))$ additive error with constant multiplicative approximation.

Both algorithms surpass previous $ ilde{O}(n^{1/2 + a})$ additive error bounds, approaching linear dependence on $n$.

Abstract

Given a data set of size $n$ in $d'$-dimensional Euclidean space, the $k$-means problem asks for a set of $k$ points (called centers) so that the sum of the $\ell_2^2$-distances between points of a given data set of size $n$ and the set of $k$ centers is minimized. Recent work on this problem in the locally private setting achieves constant multiplicative approximation with additive error $\tilde{O} (n^{1/2 + a} \cdot k \cdot \max \{\sqrt{d}, \sqrt{k} \})$ and proves a lower bound of $\Omega(\sqrt{n})$ on the additive error for any solution with a constant number of rounds. In this work we bridge the gap between the exponents of $n$ in the upper and lower bounds on the additive error with two new algorithms. Given any $\alpha>0$, our first algorithm achieves a multiplicative approximation guarantee which is at most a $(1+\alpha)$ factor greater than that of any non-private $k$-means clustering algorithm with $k^{\tilde{O}(1/\alpha^2)} \sqrt{d' n} \mbox{poly}\log n$ additive error. Given any $c>\sqrt{2}$, our second algorithm achieves $O(k^{1 + \tilde{O}(1/(2c^2-1))} \sqrt{d' n} \mbox{poly} \log n)$ additive error with constant multiplicative approximation. Both algorithms go beyond the $\Omega(n^{1/2 + a})$ factor that occurs in the additive error for arbitrarily small parameters $a$ in previous work, and the second algorithm in particular shows for the first time that it is possible to solve the locally private $k$-means problem in a constant number of rounds with constant factor multiplicative approximation and polynomial dependence on $k$ in the additive error arbitrarily close to linear.