A Faster $k$-means++ Algorithm

TL;DR

This paper introduces FastKmeans++, a new algorithm that significantly reduces the running time for initializing k-means clustering, making it nearly optimal for large datasets.

Contribution

The paper presents FastKmeans++, an algorithm that improves initialization time for k-means++ from O(n d k^2) to O(n d + n k^2), enhancing efficiency for large-scale data.

Findings

FastKmeans++ runs in O(n d + n k^2) time

Achieves nearly optimal running time for k-means++ initialization

Significantly faster than previous algorithms for large datasets

Abstract

$k$-means++ is an important algorithm for choosing initial cluster centers for the $k$-means clustering algorithm. In this work, we present a new algorithm that can solve the $k$-means++ problem with nearly optimal running time. Given $n$ data points in $\mathbb{R}^d$, the current state-of-the-art algorithm runs in $\widetilde{O}(k )$ iterations, and each iteration takes $\widetilde{O}(nd k)$ time. The overall running time is thus $\widetilde{O}(n d k^2)$. We propose a new algorithm \textsc{FastKmeans++} that only takes in $\widetilde{O}(nd + nk^2)$ time, in total.