Local hypergraph clustering using capacity releasing diffusion.

Local graph clustering is an important machine learning task that aims to find a well-connected cluster near a set of seed nodes. Recent results have revealed that incorporating higher order information significantly enhances the results of graph clustering techniques. The majority of existing resea...

Full description

Bibliographic Details
Main Authors: Rania Ibrahim, David F Gleich
Format: Article
Language:English
Published: Public Library of Science (PLoS) 2020-01-01
Series:PLoS ONE
Online Access:https://doi.org/10.1371/journal.pone.0243485
id doaj-5ad3c5200b304d38a151d59db9b6bea9
record_format Article
spelling doaj-5ad3c5200b304d38a151d59db9b6bea92021-03-04T12:53:49ZengPublic Library of Science (PLoS)PLoS ONE1932-62032020-01-011512e024348510.1371/journal.pone.0243485Local hypergraph clustering using capacity releasing diffusion.Rania IbrahimDavid F GleichLocal graph clustering is an important machine learning task that aims to find a well-connected cluster near a set of seed nodes. Recent results have revealed that incorporating higher order information significantly enhances the results of graph clustering techniques. The majority of existing research in this area focuses on spectral graph theory-based techniques. However, an alternative perspective on local graph clustering arises from using max-flow and min-cut on the objectives, which offer distinctly different guarantees. For instance, a new method called capacity releasing diffusion (CRD) was recently proposed and shown to preserve local structure around the seeds better than spectral methods. The method was also the first local clustering technique that is not subject to the quadratic Cheeger inequality by assuming a good cluster near the seed nodes. In this paper, we propose a local hypergraph clustering technique called hypergraph CRD (HG-CRD) by extending the CRD process to cluster based on higher order patterns, encoded as hyperedges of a hypergraph. Moreover, we theoretically show that HG-CRD gives results about a quantity called motif conductance, rather than a biased version used in previous experiments. Experimental results on synthetic datasets and real world graphs show that HG-CRD enhances the clustering quality.https://doi.org/10.1371/journal.pone.0243485
collection DOAJ
language English
format Article
sources DOAJ
author Rania Ibrahim
David F Gleich
spellingShingle Rania Ibrahim
David F Gleich
Local hypergraph clustering using capacity releasing diffusion.
PLoS ONE
author_facet Rania Ibrahim
David F Gleich
author_sort Rania Ibrahim
title Local hypergraph clustering using capacity releasing diffusion.
title_short Local hypergraph clustering using capacity releasing diffusion.
title_full Local hypergraph clustering using capacity releasing diffusion.
title_fullStr Local hypergraph clustering using capacity releasing diffusion.
title_full_unstemmed Local hypergraph clustering using capacity releasing diffusion.
title_sort local hypergraph clustering using capacity releasing diffusion.
publisher Public Library of Science (PLoS)
series PLoS ONE
issn 1932-6203
publishDate 2020-01-01
description Local graph clustering is an important machine learning task that aims to find a well-connected cluster near a set of seed nodes. Recent results have revealed that incorporating higher order information significantly enhances the results of graph clustering techniques. The majority of existing research in this area focuses on spectral graph theory-based techniques. However, an alternative perspective on local graph clustering arises from using max-flow and min-cut on the objectives, which offer distinctly different guarantees. For instance, a new method called capacity releasing diffusion (CRD) was recently proposed and shown to preserve local structure around the seeds better than spectral methods. The method was also the first local clustering technique that is not subject to the quadratic Cheeger inequality by assuming a good cluster near the seed nodes. In this paper, we propose a local hypergraph clustering technique called hypergraph CRD (HG-CRD) by extending the CRD process to cluster based on higher order patterns, encoded as hyperedges of a hypergraph. Moreover, we theoretically show that HG-CRD gives results about a quantity called motif conductance, rather than a biased version used in previous experiments. Experimental results on synthetic datasets and real world graphs show that HG-CRD enhances the clustering quality.
url https://doi.org/10.1371/journal.pone.0243485
work_keys_str_mv AT raniaibrahim localhypergraphclusteringusingcapacityreleasingdiffusion
AT davidfgleich localhypergraphclusteringusingcapacityreleasingdiffusion
_version_ 1714801074945130496