Measuring the stability of spectral clustering

As an indicator of the stability of spectral clustering of an undirected weighted graph into $k$ clusters, the $k$th spectral gap of the graph Laplacian is often considered. The $k$th spectral gap is characterized in this paper as an unstructured distance to ambiguity, namely as the minimal distance...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Andreotti, Eleonora, Edelmann, Dominik, Guglielmi, Nicola, Lubich, Christian
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title
container_volume
creator Andreotti, Eleonora
Edelmann, Dominik
Guglielmi, Nicola
Lubich, Christian
description As an indicator of the stability of spectral clustering of an undirected weighted graph into $k$ clusters, the $k$th spectral gap of the graph Laplacian is often considered. The $k$th spectral gap is characterized in this paper as an unstructured distance to ambiguity, namely as the minimal distance of the Laplacian to arbitrary symmetric matrices with vanishing $k$th spectral gap. As a conceptually more appropriate measure of stability, the structured distance to ambiguity of the $k$-clustering is introduced as the minimal distance of the Laplacian to Laplacians of graphs with the same vertices and edges but with weights that are perturbed such that the $k$th spectral gap vanishes. To compute a solution to this matrix nearness problem, a two-level iterative algorithm is proposed that uses a constrained gradient system of matrix differential equations in the inner iteration and a one-dimensional optimization of the perturbation size in the outer iteration. The structured and unstructured distances to ambiguity are compared on some example graphs. The numerical experiments show, in particular, that selecting the number $k$ of clusters according to the criterion of maximal stability can lead to different results for the structured and unstructured stability indicators.
doi_str_mv 10.48550/arxiv.1903.05193
format Article
fullrecord <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_1903_05193</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>1903_05193</sourcerecordid><originalsourceid>FETCH-LOGICAL-a673-88d51055f635db7544d4d43f5cae5c89f3bd520f36a8e9b0ef6df2b1be13ceae3</originalsourceid><addsrcrecordid>eNotzrFuAjEQBFA3KSKSD0iFf-AuNnt7-EqEAkEioqE_re01WLoEZBsEf58A0RTTjEZPiDet6sYgqndKl3iudaegVqg7eBb1F1M-pfizk2XPMheycYjlKg9B5iO7kmiQbjjlwrfRi3gKNGR-_e-R2C4-tvPPar1ZruazdUXtFCpjPGqFGFpAb6fYNP4vENARozNdAOtxogK0ZLizikPrw8RqyxocE8NIjB-3d3B_TPGb0rW_wfs7HH4B_2A-fQ</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Measuring the stability of spectral clustering</title><source>arXiv.org</source><creator>Andreotti, Eleonora ; Edelmann, Dominik ; Guglielmi, Nicola ; Lubich, Christian</creator><creatorcontrib>Andreotti, Eleonora ; Edelmann, Dominik ; Guglielmi, Nicola ; Lubich, Christian</creatorcontrib><description>As an indicator of the stability of spectral clustering of an undirected weighted graph into $k$ clusters, the $k$th spectral gap of the graph Laplacian is often considered. The $k$th spectral gap is characterized in this paper as an unstructured distance to ambiguity, namely as the minimal distance of the Laplacian to arbitrary symmetric matrices with vanishing $k$th spectral gap. As a conceptually more appropriate measure of stability, the structured distance to ambiguity of the $k$-clustering is introduced as the minimal distance of the Laplacian to Laplacians of graphs with the same vertices and edges but with weights that are perturbed such that the $k$th spectral gap vanishes. To compute a solution to this matrix nearness problem, a two-level iterative algorithm is proposed that uses a constrained gradient system of matrix differential equations in the inner iteration and a one-dimensional optimization of the perturbation size in the outer iteration. The structured and unstructured distances to ambiguity are compared on some example graphs. The numerical experiments show, in particular, that selecting the number $k$ of clusters according to the criterion of maximal stability can lead to different results for the structured and unstructured stability indicators.</description><identifier>DOI: 10.48550/arxiv.1903.05193</identifier><language>eng</language><subject>Computer Science - Numerical Analysis ; Mathematics - Numerical Analysis</subject><creationdate>2019-03</creationdate><rights>http://arxiv.org/licenses/nonexclusive-distrib/1.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,776,881</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/1903.05193$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.1903.05193$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Andreotti, Eleonora</creatorcontrib><creatorcontrib>Edelmann, Dominik</creatorcontrib><creatorcontrib>Guglielmi, Nicola</creatorcontrib><creatorcontrib>Lubich, Christian</creatorcontrib><title>Measuring the stability of spectral clustering</title><description>As an indicator of the stability of spectral clustering of an undirected weighted graph into $k$ clusters, the $k$th spectral gap of the graph Laplacian is often considered. The $k$th spectral gap is characterized in this paper as an unstructured distance to ambiguity, namely as the minimal distance of the Laplacian to arbitrary symmetric matrices with vanishing $k$th spectral gap. As a conceptually more appropriate measure of stability, the structured distance to ambiguity of the $k$-clustering is introduced as the minimal distance of the Laplacian to Laplacians of graphs with the same vertices and edges but with weights that are perturbed such that the $k$th spectral gap vanishes. To compute a solution to this matrix nearness problem, a two-level iterative algorithm is proposed that uses a constrained gradient system of matrix differential equations in the inner iteration and a one-dimensional optimization of the perturbation size in the outer iteration. The structured and unstructured distances to ambiguity are compared on some example graphs. The numerical experiments show, in particular, that selecting the number $k$ of clusters according to the criterion of maximal stability can lead to different results for the structured and unstructured stability indicators.</description><subject>Computer Science - Numerical Analysis</subject><subject>Mathematics - Numerical Analysis</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2019</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNotzrFuAjEQBFA3KSKSD0iFf-AuNnt7-EqEAkEioqE_re01WLoEZBsEf58A0RTTjEZPiDet6sYgqndKl3iudaegVqg7eBb1F1M-pfizk2XPMheycYjlKg9B5iO7kmiQbjjlwrfRi3gKNGR-_e-R2C4-tvPPar1ZruazdUXtFCpjPGqFGFpAb6fYNP4vENARozNdAOtxogK0ZLizikPrw8RqyxocE8NIjB-3d3B_TPGb0rW_wfs7HH4B_2A-fQ</recordid><startdate>20190312</startdate><enddate>20190312</enddate><creator>Andreotti, Eleonora</creator><creator>Edelmann, Dominik</creator><creator>Guglielmi, Nicola</creator><creator>Lubich, Christian</creator><scope>AKY</scope><scope>AKZ</scope><scope>GOX</scope></search><sort><creationdate>20190312</creationdate><title>Measuring the stability of spectral clustering</title><author>Andreotti, Eleonora ; Edelmann, Dominik ; Guglielmi, Nicola ; Lubich, Christian</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a673-88d51055f635db7544d4d43f5cae5c89f3bd520f36a8e9b0ef6df2b1be13ceae3</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2019</creationdate><topic>Computer Science - Numerical Analysis</topic><topic>Mathematics - Numerical Analysis</topic><toplevel>online_resources</toplevel><creatorcontrib>Andreotti, Eleonora</creatorcontrib><creatorcontrib>Edelmann, Dominik</creatorcontrib><creatorcontrib>Guglielmi, Nicola</creatorcontrib><creatorcontrib>Lubich, Christian</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv Mathematics</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Andreotti, Eleonora</au><au>Edelmann, Dominik</au><au>Guglielmi, Nicola</au><au>Lubich, Christian</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Measuring the stability of spectral clustering</atitle><date>2019-03-12</date><risdate>2019</risdate><abstract>As an indicator of the stability of spectral clustering of an undirected weighted graph into $k$ clusters, the $k$th spectral gap of the graph Laplacian is often considered. The $k$th spectral gap is characterized in this paper as an unstructured distance to ambiguity, namely as the minimal distance of the Laplacian to arbitrary symmetric matrices with vanishing $k$th spectral gap. As a conceptually more appropriate measure of stability, the structured distance to ambiguity of the $k$-clustering is introduced as the minimal distance of the Laplacian to Laplacians of graphs with the same vertices and edges but with weights that are perturbed such that the $k$th spectral gap vanishes. To compute a solution to this matrix nearness problem, a two-level iterative algorithm is proposed that uses a constrained gradient system of matrix differential equations in the inner iteration and a one-dimensional optimization of the perturbation size in the outer iteration. The structured and unstructured distances to ambiguity are compared on some example graphs. The numerical experiments show, in particular, that selecting the number $k$ of clusters according to the criterion of maximal stability can lead to different results for the structured and unstructured stability indicators.</abstract><doi>10.48550/arxiv.1903.05193</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext_linktorsrc
identifier DOI: 10.48550/arxiv.1903.05193
ispartof
issn
language eng
recordid cdi_arxiv_primary_1903_05193
source arXiv.org
subjects Computer Science - Numerical Analysis
Mathematics - Numerical Analysis
title Measuring the stability of spectral clustering
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-27T23%3A47%3A29IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-arxiv_GOX&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Measuring%20the%20stability%20of%20spectral%20clustering&rft.au=Andreotti,%20Eleonora&rft.date=2019-03-12&rft_id=info:doi/10.48550/arxiv.1903.05193&rft_dat=%3Carxiv_GOX%3E1903_05193%3C/arxiv_GOX%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true