Simultaneous Unsupervised Learning of Disparate Clusterings
Identifieur interne : 001143 ( Main/Exploration ); précédent : 001142; suivant : 001144Simultaneous Unsupervised Learning of Disparate Clusterings
Auteurs : Prateek Jain [États-Unis] ; Raghu Meka [États-Unis] ; Inderjit S. Dhillon [États-Unis]Source :
- Statistical Analysis and Data Mining [ 1932-1864 ] ; 2008-11-25.
English descriptors
Abstract
Most clustering algorithms produce a single clustering for a given dataset even when the data can be clustered naturally in multiple ways. In this paper, we address the difficult problem of uncovering disparate clusterings from the data in a totally unsupervised manner. We propose two new approaches for this problem. In the first approach, we aim to find good clusterings of the data that are also decorrelated with one another. To this end, we give a new and tractable characterization of decorrelation between clusterings, and present an objective function to capture it. We provide an iterative “decorrelated” k‐means type algorithm to minimize this objective function. In the second approach, we model the data as a sum of mixtures and associate each mixture with a clustering. This approach leads us to the problem of learning a convolution of mixture distributions. Though the latter problem can be formulated as one of factorial learning 8, 13, 16, the existing formulations and methods do not perform well on many real high‐dimensional datasets. We propose a new regularized factorial‐learning framework that is more suitable for capturing the notion of disparate clusterings in modern, high‐dimensional datasets. Furthermore, we provide kernelized version of both of our algorithms. The resulting algorithms do well in uncovering multiple clusterings, and are much improved over existing methods. We evaluate our methods on two real‐world datasets—a music dataset from the text‐mining domain, and a portrait dataset from the computer‐vision domain. Our methods achieve a substantially higher accuracy than existing factorial learning as well as traditional clustering algorithms. Copyright © 2008 Wiley Periodicals, Inc. Statistical Analysis and Data Mining 1: 000‐000, 2008
Url:
DOI: 10.1002/sam.10007
Affiliations:
Links toward previous steps (curation, corpus...)
- to stream Istex, to step Corpus: 002D13
- to stream Istex, to step Curation: 002852
- to stream Istex, to step Checkpoint: 000B72
- to stream Main, to step Merge: 001156
- to stream Main, to step Curation: 001143
Le document en format XML
<record><TEI wicri:istexFullTextTei="biblStruct"><teiHeader><fileDesc><titleStmt><title xml:lang="en">Simultaneous Unsupervised Learning of Disparate Clusterings</title>
<author><name sortKey="Jain, Prateek" sort="Jain, Prateek" uniqKey="Jain P" first="Prateek" last="Jain">Prateek Jain</name>
</author>
<author><name sortKey="Meka, Raghu" sort="Meka, Raghu" uniqKey="Meka R" first="Raghu" last="Meka">Raghu Meka</name>
</author>
<author><name sortKey="Dhillon, Inderjit S" sort="Dhillon, Inderjit S" uniqKey="Dhillon I" first="Inderjit S." last="Dhillon">Inderjit S. Dhillon</name>
</author>
</titleStmt>
<publicationStmt><idno type="wicri:source">ISTEX</idno>
<idno type="RBID">ISTEX:82B51EB4684DA3F88BC9E5B2A21E4F6ABB5A0A02</idno>
<date when="2008" year="2008">2008</date>
<idno type="doi">10.1002/sam.10007</idno>
<idno type="url">https://api.istex.fr/document/82B51EB4684DA3F88BC9E5B2A21E4F6ABB5A0A02/fulltext/pdf</idno>
<idno type="wicri:Area/Istex/Corpus">002D13</idno>
<idno type="wicri:Area/Istex/Curation">002852</idno>
<idno type="wicri:Area/Istex/Checkpoint">000B72</idno>
<idno type="wicri:doubleKey">1932-1864:2008:Jain P:simultaneous:unsupervised:learning</idno>
<idno type="wicri:Area/Main/Merge">001156</idno>
<idno type="wicri:Area/Main/Curation">001143</idno>
<idno type="wicri:Area/Main/Exploration">001143</idno>
</publicationStmt>
<sourceDesc><biblStruct><analytic><title level="a" type="main" xml:lang="en">Simultaneous Unsupervised Learning of Disparate Clusterings</title>
<author><name sortKey="Jain, Prateek" sort="Jain, Prateek" uniqKey="Jain P" first="Prateek" last="Jain">Prateek Jain</name>
<affiliation wicri:level="2"><country xml:lang="fr">États-Unis</country>
<wicri:regionArea>Department of Computer Sciences, University of Texas, Austin, TX 78712‐1188</wicri:regionArea>
<placeName><region type="state">Texas</region>
</placeName>
</affiliation>
</author>
<author><name sortKey="Meka, Raghu" sort="Meka, Raghu" uniqKey="Meka R" first="Raghu" last="Meka">Raghu Meka</name>
<affiliation wicri:level="2"><country xml:lang="fr">États-Unis</country>
<wicri:regionArea>Department of Computer Sciences, University of Texas, Austin, TX 78712‐1188</wicri:regionArea>
<placeName><region type="state">Texas</region>
</placeName>
</affiliation>
</author>
<author><name sortKey="Dhillon, Inderjit S" sort="Dhillon, Inderjit S" uniqKey="Dhillon I" first="Inderjit S." last="Dhillon">Inderjit S. Dhillon</name>
<affiliation wicri:level="2"><country xml:lang="fr">États-Unis</country>
<wicri:regionArea>Department of Computer Sciences, University of Texas, Austin, TX 78712‐1188</wicri:regionArea>
<placeName><region type="state">Texas</region>
</placeName>
</affiliation>
</author>
</analytic>
<monogr></monogr>
<series><title level="j">Statistical Analysis and Data Mining</title>
<title level="j" type="abbrev">Statistical Analy Data Mining</title>
<idno type="ISSN">1932-1864</idno>
<idno type="eISSN">1932-1872</idno>
<imprint><publisher>Wiley Subscription Services, Inc., A Wiley Company</publisher>
<pubPlace>Hoboken</pubPlace>
<date type="published" when="2008-11-25">2008-11-25</date>
<biblScope unit="volume">1</biblScope>
<biblScope unit="issue">3</biblScope>
<biblScope unit="page" from="195">195</biblScope>
<biblScope unit="page" to="210">210</biblScope>
</imprint>
<idno type="ISSN">1932-1864</idno>
</series>
<idno type="istex">82B51EB4684DA3F88BC9E5B2A21E4F6ABB5A0A02</idno>
<idno type="DOI">10.1002/sam.10007</idno>
<idno type="ArticleID">SAM10007</idno>
</biblStruct>
</sourceDesc>
<seriesStmt><idno type="ISSN">1932-1864</idno>
</seriesStmt>
</fileDesc>
<profileDesc><textClass><keywords scheme="KwdEn" xml:lang="en"><term>disparate clustering</term>
<term>expectation maximization</term>
<term>k‐means</term>
<term>unsupervised learning</term>
</keywords>
</textClass>
<langUsage><language ident="en">en</language>
</langUsage>
</profileDesc>
</teiHeader>
<front><div type="abstract" xml:lang="en">Most clustering algorithms produce a single clustering for a given dataset even when the data can be clustered naturally in multiple ways. In this paper, we address the difficult problem of uncovering disparate clusterings from the data in a totally unsupervised manner. We propose two new approaches for this problem. In the first approach, we aim to find good clusterings of the data that are also decorrelated with one another. To this end, we give a new and tractable characterization of decorrelation between clusterings, and present an objective function to capture it. We provide an iterative “decorrelated” k‐means type algorithm to minimize this objective function. In the second approach, we model the data as a sum of mixtures and associate each mixture with a clustering. This approach leads us to the problem of learning a convolution of mixture distributions. Though the latter problem can be formulated as one of factorial learning 8, 13, 16, the existing formulations and methods do not perform well on many real high‐dimensional datasets. We propose a new regularized factorial‐learning framework that is more suitable for capturing the notion of disparate clusterings in modern, high‐dimensional datasets. Furthermore, we provide kernelized version of both of our algorithms. The resulting algorithms do well in uncovering multiple clusterings, and are much improved over existing methods. We evaluate our methods on two real‐world datasets—a music dataset from the text‐mining domain, and a portrait dataset from the computer‐vision domain. Our methods achieve a substantially higher accuracy than existing factorial learning as well as traditional clustering algorithms. Copyright © 2008 Wiley Periodicals, Inc. Statistical Analysis and Data Mining 1: 000‐000, 2008</div>
</front>
</TEI>
<affiliations><list><country><li>États-Unis</li>
</country>
<region><li>Texas</li>
</region>
</list>
<tree><country name="États-Unis"><region name="Texas"><name sortKey="Jain, Prateek" sort="Jain, Prateek" uniqKey="Jain P" first="Prateek" last="Jain">Prateek Jain</name>
</region>
<name sortKey="Dhillon, Inderjit S" sort="Dhillon, Inderjit S" uniqKey="Dhillon I" first="Inderjit S." last="Dhillon">Inderjit S. Dhillon</name>
<name sortKey="Meka, Raghu" sort="Meka, Raghu" uniqKey="Meka R" first="Raghu" last="Meka">Raghu Meka</name>
</country>
</tree>
</affiliations>
</record>
Pour manipuler ce document sous Unix (Dilib)
EXPLOR_STEP=$WICRI_ROOT/Wicri/Musique/explor/MozartV1/Data/Main/Exploration
HfdSelect -h $EXPLOR_STEP/biblio.hfd -nk 001143 | SxmlIndent | more
Ou
HfdSelect -h $EXPLOR_AREA/Data/Main/Exploration/biblio.hfd -nk 001143 | SxmlIndent | more
Pour mettre un lien sur cette page dans le réseau Wicri
{{Explor lien |wiki= Wicri/Musique |area= MozartV1 |flux= Main |étape= Exploration |type= RBID |clé= ISTEX:82B51EB4684DA3F88BC9E5B2A21E4F6ABB5A0A02 |texte= Simultaneous Unsupervised Learning of Disparate Clusterings }}
This area was generated with Dilib version V0.6.20. |