Saved in:
Bibliographic Details
Main Authors: Ostovari, Mojtaba, Zarei, Alireza
Format: Preprint
Published: 2023
Subjects:
Online Access:https://arxiv.org/abs/2310.09638
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909689501974528
author Ostovari, Mojtaba
Zarei, Alireza
author_facet Ostovari, Mojtaba
Zarei, Alireza
contents We present combinatorial approximation algorithms for the weighted correlation clustering problem. In this problem, we have a set of vertices and two weight values for each pair of vertices, denoting their difference and similarity. The goal is to cluster the vertices with minimum total intra-cluster difference weights plus inter-cluster similarity weights. We present two results for weighted instances with $n$ vertices: - A randomized 3-approximation combinatorial algorithm for instances that satisfy probability constraints, running in $O(n^2)$ time. This improves the $O(n^6)$ running time of the previous best-known combinatorial approximation, a 3-approximation algorithm, introduced by Chawla et al. (2015). - A randomized 1.6-approximation combinatorial algorithm for instances that satisfy probability and triangle inequality constraints, running in $O(n^2)$ time. This improves the longstanding combinatorial 2-approximation of Ailon et al. (2008) while matching its running time.
format Preprint
id arxiv_https___arxiv_org_abs_2310_09638
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Improved Combinatorial Approximations for Weighted Correlation Clustering
Ostovari, Mojtaba
Zarei, Alireza
Data Structures and Algorithms
We present combinatorial approximation algorithms for the weighted correlation clustering problem. In this problem, we have a set of vertices and two weight values for each pair of vertices, denoting their difference and similarity. The goal is to cluster the vertices with minimum total intra-cluster difference weights plus inter-cluster similarity weights. We present two results for weighted instances with $n$ vertices: - A randomized 3-approximation combinatorial algorithm for instances that satisfy probability constraints, running in $O(n^2)$ time. This improves the $O(n^6)$ running time of the previous best-known combinatorial approximation, a 3-approximation algorithm, introduced by Chawla et al. (2015). - A randomized 1.6-approximation combinatorial algorithm for instances that satisfy probability and triangle inequality constraints, running in $O(n^2)$ time. This improves the longstanding combinatorial 2-approximation of Ailon et al. (2008) while matching its running time.
title Improved Combinatorial Approximations for Weighted Correlation Clustering
topic Data Structures and Algorithms
url https://arxiv.org/abs/2310.09638