Linear time small coresets for k-mean clustering of segments with applications

TL;DR

This paper introduces a novel linear-time algorithm for constructing small coresets for k-means clustering of segments, enabling efficient and accurate clustering in streaming and distributed settings with practical applications.

Contribution

It presents the first coreset construction for arbitrary segments in k-means clustering, achieving size $O( ext{log}^2 n)$ in linear time for fixed parameters.

Findings

Coresets enable fast, approximate clustering with minimal accuracy loss.

Method scales efficiently to large datasets and real-time applications.

Experimental results show significant speedups with maintained accuracy.

Abstract

We study the $k$-means problem for a set $\mathcal{S} \subseteq \mathbb{R}^d$ of $n$ segments, aiming to find $k$ centers $X \subseteq \mathbb{R}^d$ that minimize $D(\mathcal{S},X) := \sum_{S \in \mathcal{S}} \min_{x \in X} D(S,x)$, where $D(S,x) := \int_{p \in S} |p - x| dp$ measures the total distance from each point along a segment to a center. Variants of this problem include handling outliers, employing alternative distance functions such as M-estimators, weighting distances to achieve balanced clustering, or enforcing unique cluster assignments. For any $\varepsilon > 0$, an $\varepsilon$-coreset is a weighted subset $C \subseteq \mathbb{R}^d$ that approximates $D(\mathcal{S},X)$ within a factor of $1 \pm \varepsilon$ for any set of $k$ centers, enabling efficient streaming, distributed, or parallel computation. We propose the first coreset construction that provably handles arbitrary input segments. For constant $k$ and $\varepsilon$, it produces a coreset of size $O(\log^2 n)$ computable in $O(nd)$ time. Experiments, including a real-time video tracking application, demonstrate substantial speedups with minimal loss in clustering accuracy, confirming both the practical efficiency and theoretical guarantees of our method.