Kernel Additive Models for Source Separation
Identifieur interne : 000E40 ( Main/Exploration ); précédent : 000E39; suivant : 000E41Kernel Additive Models for Source Separation
Auteurs : Antoine Liutkus [France] ; Derry Fitzgerald [Irlande (pays)] ; Zafar Rafii [États-Unis] ; Bryan Pardo [États-Unis] ; Laurent Daudet [France]Source :
- IEEE transactions on signal processing [ 1053-587X ] ; 2014.
Descripteurs français
- Pascal (Inist)
English descriptors
- KwdEn :
- mix :
Abstract
Source separation consists of separating a signal into additive components. It is a topic of considerable interest with many applications that has gathered much attention recently. Here, we introduce a new framework for source separation called Kernel Additive Modelling, which is based on local regression and permits efficient separation of multidimensional and/or non-negative and/or non-regularly sampled signals. The main idea of the method is to assume that a source at some location can be estimated using its values at other locations nearby, where nearness is defined through a source-specific proximity kernel. Such a kernel provides an efficient way to account for features like periodicity, continuity, smoothness, stability over time or frequency, and self-similarity. In many cases, such local dynamics are indeed much more natural to assess than any global model such as a tensor factorization. This framework permits one to use different proximity kernels for different sources and to separate them using the iterative kernel backfitting algorithm we describe. As we show, kernel additive modelling generalizes many recent and efficient techniques for source separation and opens the path to creating and combining source models in a principled way. Experimental results on the separation of synthetic and audio signals demonstrate the effectiveness of the approach.
Url:
Affiliations:
- France, Irlande (pays), États-Unis
- Grand Est, Lorraine (région), Île-de-France
- Université de Lorraine
Links toward previous steps (curation, corpus...)
- to stream PascalFrancis, to step Corpus: 000013
- to stream PascalFrancis, to step Curation: 000991
- to stream PascalFrancis, to step Checkpoint: 000015
- to stream Main, to step Merge: 000E32
- to stream Hal, to step Corpus: 000023
- to stream Hal, to step Curation: 000023
- to stream Hal, to step Checkpoint: 000A81
- to stream Main, to step Merge: 000B40
- to stream Main, to step Curation: 000E40
Le document en format XML
<record><TEI><teiHeader><fileDesc><titleStmt><title xml:lang="en" level="a">Kernel Additive Models for Source Separation</title>
<author><name sortKey="Liutkus, Antoine" sort="Liutkus, Antoine" uniqKey="Liutkus A" first="Antoine" last="Liutkus">Antoine Liutkus</name>
<affiliation wicri:level="3"><inist:fA14 i1="01"><s1>Inria, Villers-lès-Nancy, 54600</s1>
<s3>FRA</s3>
<sZ>1 aut.</sZ>
</inist:fA14>
<country>France</country>
<placeName><region type="region" nuts="2">Grand Est</region>
<region type="old region" nuts="2">Lorraine (région)</region>
</placeName>
</affiliation>
<affiliation wicri:level="4"><inist:fA14 i1="02"><s1>Université de Lorraine, LORIA, UMR 7503</s1>
<s2>Villers-lès-Nancy, 54600</s2>
<s3>FRA</s3>
<sZ>1 aut.</sZ>
</inist:fA14>
<country>France</country>
<placeName><region type="region" nuts="2">Grand Est</region>
<region type="old region" nuts="2">Lorraine (région)</region>
</placeName>
<orgName type="university">Université de Lorraine</orgName>
</affiliation>
<affiliation wicri:level="3"><inist:fA14 i1="03"><s1>CNRS, LORIA, UMR 7503</s1>
<s2>Villers-lès-Nancy, 54600</s2>
<s3>FRA</s3>
<sZ>1 aut.</sZ>
</inist:fA14>
<country>France</country>
<placeName><region type="region" nuts="2">Grand Est</region>
<region type="old region" nuts="2">Lorraine (région)</region>
</placeName>
</affiliation>
</author>
<author><name sortKey="Fitzgerald, Derry" sort="Fitzgerald, Derry" uniqKey="Fitzgerald D" first="Derry" last="Fitzgerald">Derry Fitzgerald</name>
<affiliation wicri:level="1"><inist:fA14 i1="04"><s1>NIMBUS Centre, Cork Institute of Technology</s1>
<s2>Cork</s2>
<s3>IRL</s3>
<sZ>2 aut.</sZ>
</inist:fA14>
<country>Irlande (pays)</country>
<wicri:noRegion>Cork</wicri:noRegion>
</affiliation>
</author>
<author><name sortKey="Rafii, Zafar" sort="Rafii, Zafar" uniqKey="Rafii Z" first="Zafar" last="Rafii">Zafar Rafii</name>
<affiliation wicri:level="1"><inist:fA14 i1="05"><s1>Northwestern University</s1>
<s2>Evanston, IL 60208</s2>
<s3>USA</s3>
<sZ>3 aut.</sZ>
<sZ>4 aut.</sZ>
</inist:fA14>
<country>États-Unis</country>
<wicri:noRegion>Northwestern University</wicri:noRegion>
</affiliation>
</author>
<author><name sortKey="Pardo, Bryan" sort="Pardo, Bryan" uniqKey="Pardo B" first="Bryan" last="Pardo">Bryan Pardo</name>
<affiliation wicri:level="1"><inist:fA14 i1="05"><s1>Northwestern University</s1>
<s2>Evanston, IL 60208</s2>
<s3>USA</s3>
<sZ>3 aut.</sZ>
<sZ>4 aut.</sZ>
</inist:fA14>
<country>États-Unis</country>
<wicri:noRegion>Northwestern University</wicri:noRegion>
</affiliation>
</author>
<author><name sortKey="Daudet, Laurent" sort="Daudet, Laurent" uniqKey="Daudet L" first="Laurent" last="Daudet">Laurent Daudet</name>
<affiliation wicri:level="3"><inist:fA14 i1="06"><s1>Institut Langevin, Paris Diderot University</s1>
<s2>Paris 75005</s2>
<s3>FRA</s3>
<sZ>5 aut.</sZ>
</inist:fA14>
<country>France</country>
<placeName><region type="region" nuts="2">Île-de-France</region>
</placeName>
</affiliation>
</author>
</titleStmt>
<publicationStmt><idno type="wicri:source">INIST</idno>
<idno type="inist">14-0219788</idno>
<date when="2014">2014</date>
<idno type="stanalyst">PASCAL 14-0219788 INIST</idno>
<idno type="RBID">Pascal:14-0219788</idno>
<idno type="wicri:Area/PascalFrancis/Corpus">000013</idno>
<idno type="wicri:Area/PascalFrancis/Curation">000991</idno>
<idno type="wicri:Area/PascalFrancis/Checkpoint">000015</idno>
<idno type="wicri:explorRef" wicri:stream="PascalFrancis" wicri:step="Checkpoint">000015</idno>
<idno type="wicri:doubleKey">1053-587X:2014:Liutkus A:kernel:additive:models</idno>
<idno type="wicri:Area/Main/Merge">000E32</idno>
<idno type="wicri:source">HAL</idno>
<idno type="RBID">Hal:hal-01011044</idno>
<idno type="url">https://hal.inria.fr/hal-01011044</idno>
<idno type="wicri:Area/Hal/Corpus">000023</idno>
<idno type="wicri:Area/Hal/Curation">000023</idno>
<idno type="wicri:Area/Hal/Checkpoint">000A81</idno>
<idno type="wicri:explorRef" wicri:stream="Hal" wicri:step="Checkpoint">000A81</idno>
<idno type="wicri:doubleKey">1053-587X:2014:Liutkus A:kernel:additive:models</idno>
<idno type="wicri:Area/Main/Merge">000B40</idno>
<idno type="wicri:Area/Main/Curation">000E40</idno>
<idno type="wicri:Area/Main/Exploration">000E40</idno>
</publicationStmt>
<sourceDesc><biblStruct><analytic><title xml:lang="en" level="a">Kernel Additive Models for Source Separation</title>
<author><name sortKey="Liutkus, Antoine" sort="Liutkus, Antoine" uniqKey="Liutkus A" first="Antoine" last="Liutkus">Antoine Liutkus</name>
<affiliation wicri:level="3"><inist:fA14 i1="01"><s1>Inria, Villers-lès-Nancy, 54600</s1>
<s3>FRA</s3>
<sZ>1 aut.</sZ>
</inist:fA14>
<country>France</country>
<placeName><region type="region" nuts="2">Grand Est</region>
<region type="old region" nuts="2">Lorraine (région)</region>
</placeName>
</affiliation>
<affiliation wicri:level="4"><inist:fA14 i1="02"><s1>Université de Lorraine, LORIA, UMR 7503</s1>
<s2>Villers-lès-Nancy, 54600</s2>
<s3>FRA</s3>
<sZ>1 aut.</sZ>
</inist:fA14>
<country>France</country>
<placeName><region type="region" nuts="2">Grand Est</region>
<region type="old region" nuts="2">Lorraine (région)</region>
</placeName>
<orgName type="university">Université de Lorraine</orgName>
</affiliation>
<affiliation wicri:level="3"><inist:fA14 i1="03"><s1>CNRS, LORIA, UMR 7503</s1>
<s2>Villers-lès-Nancy, 54600</s2>
<s3>FRA</s3>
<sZ>1 aut.</sZ>
</inist:fA14>
<country>France</country>
<placeName><region type="region" nuts="2">Grand Est</region>
<region type="old region" nuts="2">Lorraine (région)</region>
</placeName>
</affiliation>
</author>
<author><name sortKey="Fitzgerald, Derry" sort="Fitzgerald, Derry" uniqKey="Fitzgerald D" first="Derry" last="Fitzgerald">Derry Fitzgerald</name>
<affiliation wicri:level="1"><inist:fA14 i1="04"><s1>NIMBUS Centre, Cork Institute of Technology</s1>
<s2>Cork</s2>
<s3>IRL</s3>
<sZ>2 aut.</sZ>
</inist:fA14>
<country>Irlande (pays)</country>
<wicri:noRegion>Cork</wicri:noRegion>
</affiliation>
</author>
<author><name sortKey="Rafii, Zafar" sort="Rafii, Zafar" uniqKey="Rafii Z" first="Zafar" last="Rafii">Zafar Rafii</name>
<affiliation wicri:level="1"><inist:fA14 i1="05"><s1>Northwestern University</s1>
<s2>Evanston, IL 60208</s2>
<s3>USA</s3>
<sZ>3 aut.</sZ>
<sZ>4 aut.</sZ>
</inist:fA14>
<country>États-Unis</country>
<wicri:noRegion>Northwestern University</wicri:noRegion>
</affiliation>
</author>
<author><name sortKey="Pardo, Bryan" sort="Pardo, Bryan" uniqKey="Pardo B" first="Bryan" last="Pardo">Bryan Pardo</name>
<affiliation wicri:level="1"><inist:fA14 i1="05"><s1>Northwestern University</s1>
<s2>Evanston, IL 60208</s2>
<s3>USA</s3>
<sZ>3 aut.</sZ>
<sZ>4 aut.</sZ>
</inist:fA14>
<country>États-Unis</country>
<wicri:noRegion>Northwestern University</wicri:noRegion>
</affiliation>
</author>
<author><name sortKey="Daudet, Laurent" sort="Daudet, Laurent" uniqKey="Daudet L" first="Laurent" last="Daudet">Laurent Daudet</name>
<affiliation wicri:level="3"><inist:fA14 i1="06"><s1>Institut Langevin, Paris Diderot University</s1>
<s2>Paris 75005</s2>
<s3>FRA</s3>
<sZ>5 aut.</sZ>
</inist:fA14>
<country>France</country>
<placeName><region type="region" nuts="2">Île-de-France</region>
</placeName>
</affiliation>
</author>
</analytic>
<series><title level="j" type="main">IEEE transactions on signal processing</title>
<title level="j" type="abbreviated">IEEE trans. signal process.</title>
<idno type="ISSN">1053-587X</idno>
<imprint><date when="2014">2014</date>
</imprint>
</series>
</biblStruct>
</sourceDesc>
<seriesStmt><title level="j" type="main">IEEE transactions on signal processing</title>
<title level="j" type="abbreviated">IEEE trans. signal process.</title>
<idno type="ISSN">1053-587X</idno>
</seriesStmt>
</fileDesc>
<profileDesc><textClass><keywords scheme="KwdEn" xml:lang="en"><term>Algorithm</term>
<term>Audio signal</term>
<term>Factorization</term>
<term>Iterative method</term>
<term>Kernel method</term>
<term>Modeling</term>
<term>Parameter estimation</term>
<term>Performance evaluation</term>
<term>Sampled signal</term>
<term>Selfsimilarity</term>
<term>Signal estimation</term>
<term>Signal processing</term>
<term>Source localization</term>
<term>Source separation</term>
<term>Target detection</term>
</keywords>
<keywords scheme="Pascal" xml:lang="fr"><term>Méthode noyau</term>
<term>Séparation source</term>
<term>Estimation paramètre</term>
<term>Modélisation</term>
<term>Signal échantillonné</term>
<term>Détection cible</term>
<term>Localisation source</term>
<term>Autosimilitude</term>
<term>Factorisation</term>
<term>Méthode itérative</term>
<term>Algorithme</term>
<term>Signal audio</term>
<term>Evaluation performance</term>
<term>Traitement signal</term>
<term>Estimation signal</term>
</keywords>
<keywords scheme="mix" xml:lang="en"><term>audio</term>
<term>kernel models</term>
<term>machine learning</term>
<term>source separation</term>
</keywords>
</textClass>
</profileDesc>
</teiHeader>
<front><div type="abstract" xml:lang="en">Source separation consists of separating a signal into additive components. It is a topic of considerable interest with many applications that has gathered much attention recently. Here, we introduce a new framework for source separation called Kernel Additive Modelling, which is based on local regression and permits efficient separation of multidimensional and/or non-negative and/or non-regularly sampled signals. The main idea of the method is to assume that a source at some location can be estimated using its values at other locations nearby, where nearness is defined through a source-specific proximity kernel. Such a kernel provides an efficient way to account for features like periodicity, continuity, smoothness, stability over time or frequency, and self-similarity. In many cases, such local dynamics are indeed much more natural to assess than any global model such as a tensor factorization. This framework permits one to use different proximity kernels for different sources and to separate them using the iterative kernel backfitting algorithm we describe. As we show, kernel additive modelling generalizes many recent and efficient techniques for source separation and opens the path to creating and combining source models in a principled way. Experimental results on the separation of synthetic and audio signals demonstrate the effectiveness of the approach.</div>
</front>
</TEI>
<affiliations><list><country><li>France</li>
<li>Irlande (pays)</li>
<li>États-Unis</li>
</country>
<region><li>Grand Est</li>
<li>Lorraine (région)</li>
<li>Île-de-France</li>
</region>
<orgName><li>Université de Lorraine</li>
</orgName>
</list>
<tree><country name="France"><region name="Grand Est"><name sortKey="Liutkus, Antoine" sort="Liutkus, Antoine" uniqKey="Liutkus A" first="Antoine" last="Liutkus">Antoine Liutkus</name>
</region>
<name sortKey="Daudet, Laurent" sort="Daudet, Laurent" uniqKey="Daudet L" first="Laurent" last="Daudet">Laurent Daudet</name>
<name sortKey="Liutkus, Antoine" sort="Liutkus, Antoine" uniqKey="Liutkus A" first="Antoine" last="Liutkus">Antoine Liutkus</name>
<name sortKey="Liutkus, Antoine" sort="Liutkus, Antoine" uniqKey="Liutkus A" first="Antoine" last="Liutkus">Antoine Liutkus</name>
</country>
<country name="Irlande (pays)"><noRegion><name sortKey="Fitzgerald, Derry" sort="Fitzgerald, Derry" uniqKey="Fitzgerald D" first="Derry" last="Fitzgerald">Derry Fitzgerald</name>
</noRegion>
</country>
<country name="États-Unis"><noRegion><name sortKey="Rafii, Zafar" sort="Rafii, Zafar" uniqKey="Rafii Z" first="Zafar" last="Rafii">Zafar Rafii</name>
</noRegion>
<name sortKey="Pardo, Bryan" sort="Pardo, Bryan" uniqKey="Pardo B" first="Bryan" last="Pardo">Bryan Pardo</name>
</country>
</tree>
</affiliations>
</record>
Pour manipuler ce document sous Unix (Dilib)
EXPLOR_STEP=$WICRI_ROOT/Wicri/Lorraine/explor/InforLorV4/Data/Main/Exploration
HfdSelect -h $EXPLOR_STEP/biblio.hfd -nk 000E40 | SxmlIndent | more
Ou
HfdSelect -h $EXPLOR_AREA/Data/Main/Exploration/biblio.hfd -nk 000E40 | SxmlIndent | more
Pour mettre un lien sur cette page dans le réseau Wicri
{{Explor lien |wiki= Wicri/Lorraine |area= InforLorV4 |flux= Main |étape= Exploration |type= RBID |clé= Pascal:14-0219788 |texte= Kernel Additive Models for Source Separation }}
![]() | This area was generated with Dilib version V0.6.33. | ![]() |