Classification and active learning from evolving data streams in the presence of uncertain labeling knowledge
Identifieur interne : 001474 ( Hal/Curation ); précédent : 001473; suivant : 001475Classification and active learning from evolving data streams in the presence of uncertain labeling knowledge
Auteurs : Mohamed-Rafik Bouguelia [France]Source :
Descripteurs français
- mix :
- Wicri :
- topic : Classification.
English descriptors
Abstract
This thesis focuses on machine learning for data classification. To reduce the labelling cost, active learning allows to query the class label of only some important instances from a human labeller. We propose a new uncertainty measure that characterizes the importance of data and improves the performance of active learning compared to the existing uncertainty measures. This measure determines the smallest instance weight to associate with new data, so that the classifier changes its prediction concerning this data. We then consider a setting where the data arrives continuously from an infinite length stream. We propose an adaptive uncertainty threshold that is suitable for active learning in the streaming setting and achieves a compromise between the number of classification errors and the number of required labels. The existing stream-based active learning methods are initialized with some labelled instances that cover all possible classes. However, in many applications, the evolving nature of the stream implies that new classes can appear at any time. We propose an effective method of active detection of novel classes in a multi-class data stream. This method incrementally maintains a feature space area which is covered by the known classes, and detects those instances that are self-similar and external to that area as novel classes. Finally, it is often difficult to get a completely reliable labelling because the human labeller is subject to labelling errors that reduce the performance of the learned classifier. This problem was solved by introducing a measure that reflects the degree of disagreement between the manually given class and the predicted class, and a new informativeness measure that expresses the necessity for a mislabelled instance to be re-labeled by an alternative labeller.
Url:
Links toward previous steps (curation, corpus...)
- to stream Hal, to step Corpus: Pour aller vers cette notice dans l'étape Curation :001474
Links to Exploration step
Hal:tel-01262775Le document en format XML
<record><TEI><teiHeader><fileDesc><titleStmt><title xml:lang="en">Classification and active learning from evolving data streams in the presence of uncertain labeling knowledge</title>
<title xml:lang="fr">Classification et apprentissage actif à partir d'un flux de données évolutif en présence d'étiquetage incertain</title>
<author><name sortKey="Bouguelia, Mohamed Rafik" sort="Bouguelia, Mohamed Rafik" uniqKey="Bouguelia M" first="Mohamed-Rafik" last="Bouguelia">Mohamed-Rafik Bouguelia</name>
<affiliation wicri:level="1"><hal:affiliation type="researchteam" xml:id="struct-206042" status="VALID"><orgName>Recognition of writing and analysis of documents</orgName>
<orgName type="acronym">READ</orgName>
<desc><address><country key="FR"></country>
</address>
</desc>
<listRelation><relation active="#struct-423086" type="direct"></relation>
<relation active="#struct-206040" type="indirect"></relation>
<relation active="#struct-300009" type="indirect"></relation>
<relation active="#struct-413289" type="indirect"></relation>
<relation name="UMR7503" active="#struct-441569" type="indirect"></relation>
</listRelation>
<tutelles><tutelle active="#struct-423086" type="direct"><org type="department" xml:id="struct-423086" status="VALID"><orgName>Department of Natural Language Processing & Knowledge Discovery</orgName>
<orgName type="acronym">LORIA - NLPKD</orgName>
<desc><address><country key="FR"></country>
</address>
<ref type="url">http://www.loria.fr/la-recherche-en/departements/Knowledge-and-Language-Management</ref>
</desc>
<listRelation><relation active="#struct-206040" type="direct"></relation>
<relation active="#struct-300009" type="indirect"></relation>
<relation active="#struct-413289" type="indirect"></relation>
<relation name="UMR7503" active="#struct-441569" type="indirect"></relation>
</listRelation>
</org>
</tutelle>
<tutelle active="#struct-206040" type="indirect"><org type="laboratory" xml:id="struct-206040" status="VALID"><idno type="IdRef">067077927</idno>
<idno type="RNSR">198912571S</idno>
<idno type="IdUnivLorraine">[UL]RSI--</idno>
<orgName>Laboratoire Lorrain de Recherche en Informatique et ses Applications</orgName>
<orgName type="acronym">LORIA</orgName>
<date type="start">2012-01-01</date>
<desc><address><addrLine>Campus Scientifique BP 239 54506 Vandoeuvre-lès-Nancy Cedex</addrLine>
<country key="FR"></country>
</address>
<ref type="url">http://www.loria.fr</ref>
</desc>
<listRelation><relation active="#struct-300009" type="direct"></relation>
<relation active="#struct-413289" type="direct"></relation>
<relation name="UMR7503" active="#struct-441569" type="direct"></relation>
</listRelation>
</org>
</tutelle>
<tutelle active="#struct-300009" type="indirect"><org type="institution" xml:id="struct-300009" status="VALID"><orgName>Institut National de Recherche en Informatique et en Automatique</orgName>
<orgName type="acronym">Inria</orgName>
<desc><address><addrLine>Domaine de VoluceauRocquencourt - BP 10578153 Le Chesnay Cedex</addrLine>
<country key="FR"></country>
</address>
<ref type="url">http://www.inria.fr/en/</ref>
</desc>
</org>
</tutelle>
<tutelle active="#struct-413289" type="indirect"><org type="institution" xml:id="struct-413289" status="VALID"><idno type="IdRef">157040569</idno>
<idno type="IdUnivLorraine">[UL]100--</idno>
<orgName>Université de Lorraine</orgName>
<orgName type="acronym">UL</orgName>
<date type="start">2012-01-01</date>
<desc><address><addrLine>34 cours Léopold - CS 25233 - 54052 Nancy cedex</addrLine>
<country key="FR"></country>
</address>
<ref type="url">http://www.univ-lorraine.fr/</ref>
</desc>
</org>
</tutelle>
<tutelle name="UMR7503" active="#struct-441569" type="indirect"><org type="institution" xml:id="struct-441569" status="VALID"><idno type="ISNI">0000000122597504</idno>
<idno type="IdRef">02636817X</idno>
<orgName>Centre National de la Recherche Scientifique</orgName>
<orgName type="acronym">CNRS</orgName>
<date type="start">1939-10-19</date>
<desc><address><country key="FR"></country>
</address>
<ref type="url">http://www.cnrs.fr/</ref>
</desc>
</org>
</tutelle>
</tutelles>
</hal:affiliation>
<country>France</country>
<placeName><settlement type="city">Nancy</settlement>
<settlement type="city">Metz</settlement>
<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>
</author>
</titleStmt>
<publicationStmt><idno type="wicri:source">HAL</idno>
<idno type="RBID">Hal:tel-01262775</idno>
<idno type="halId">tel-01262775</idno>
<idno type="halUri">https://hal.inria.fr/tel-01262775</idno>
<idno type="url">https://hal.inria.fr/tel-01262775</idno>
<date when="2015-03-25">2015-03-25</date>
<idno type="wicri:Area/Hal/Corpus">001474</idno>
<idno type="wicri:Area/Hal/Curation">001474</idno>
</publicationStmt>
<sourceDesc><biblStruct><analytic><title xml:lang="en">Classification and active learning from evolving data streams in the presence of uncertain labeling knowledge</title>
<title xml:lang="fr">Classification et apprentissage actif à partir d'un flux de données évolutif en présence d'étiquetage incertain</title>
<author><name sortKey="Bouguelia, Mohamed Rafik" sort="Bouguelia, Mohamed Rafik" uniqKey="Bouguelia M" first="Mohamed-Rafik" last="Bouguelia">Mohamed-Rafik Bouguelia</name>
<affiliation wicri:level="1"><hal:affiliation type="researchteam" xml:id="struct-206042" status="VALID"><orgName>Recognition of writing and analysis of documents</orgName>
<orgName type="acronym">READ</orgName>
<desc><address><country key="FR"></country>
</address>
</desc>
<listRelation><relation active="#struct-423086" type="direct"></relation>
<relation active="#struct-206040" type="indirect"></relation>
<relation active="#struct-300009" type="indirect"></relation>
<relation active="#struct-413289" type="indirect"></relation>
<relation name="UMR7503" active="#struct-441569" type="indirect"></relation>
</listRelation>
<tutelles><tutelle active="#struct-423086" type="direct"><org type="department" xml:id="struct-423086" status="VALID"><orgName>Department of Natural Language Processing & Knowledge Discovery</orgName>
<orgName type="acronym">LORIA - NLPKD</orgName>
<desc><address><country key="FR"></country>
</address>
<ref type="url">http://www.loria.fr/la-recherche-en/departements/Knowledge-and-Language-Management</ref>
</desc>
<listRelation><relation active="#struct-206040" type="direct"></relation>
<relation active="#struct-300009" type="indirect"></relation>
<relation active="#struct-413289" type="indirect"></relation>
<relation name="UMR7503" active="#struct-441569" type="indirect"></relation>
</listRelation>
</org>
</tutelle>
<tutelle active="#struct-206040" type="indirect"><org type="laboratory" xml:id="struct-206040" status="VALID"><idno type="IdRef">067077927</idno>
<idno type="RNSR">198912571S</idno>
<idno type="IdUnivLorraine">[UL]RSI--</idno>
<orgName>Laboratoire Lorrain de Recherche en Informatique et ses Applications</orgName>
<orgName type="acronym">LORIA</orgName>
<date type="start">2012-01-01</date>
<desc><address><addrLine>Campus Scientifique BP 239 54506 Vandoeuvre-lès-Nancy Cedex</addrLine>
<country key="FR"></country>
</address>
<ref type="url">http://www.loria.fr</ref>
</desc>
<listRelation><relation active="#struct-300009" type="direct"></relation>
<relation active="#struct-413289" type="direct"></relation>
<relation name="UMR7503" active="#struct-441569" type="direct"></relation>
</listRelation>
</org>
</tutelle>
<tutelle active="#struct-300009" type="indirect"><org type="institution" xml:id="struct-300009" status="VALID"><orgName>Institut National de Recherche en Informatique et en Automatique</orgName>
<orgName type="acronym">Inria</orgName>
<desc><address><addrLine>Domaine de VoluceauRocquencourt - BP 10578153 Le Chesnay Cedex</addrLine>
<country key="FR"></country>
</address>
<ref type="url">http://www.inria.fr/en/</ref>
</desc>
</org>
</tutelle>
<tutelle active="#struct-413289" type="indirect"><org type="institution" xml:id="struct-413289" status="VALID"><idno type="IdRef">157040569</idno>
<idno type="IdUnivLorraine">[UL]100--</idno>
<orgName>Université de Lorraine</orgName>
<orgName type="acronym">UL</orgName>
<date type="start">2012-01-01</date>
<desc><address><addrLine>34 cours Léopold - CS 25233 - 54052 Nancy cedex</addrLine>
<country key="FR"></country>
</address>
<ref type="url">http://www.univ-lorraine.fr/</ref>
</desc>
</org>
</tutelle>
<tutelle name="UMR7503" active="#struct-441569" type="indirect"><org type="institution" xml:id="struct-441569" status="VALID"><idno type="ISNI">0000000122597504</idno>
<idno type="IdRef">02636817X</idno>
<orgName>Centre National de la Recherche Scientifique</orgName>
<orgName type="acronym">CNRS</orgName>
<date type="start">1939-10-19</date>
<desc><address><country key="FR"></country>
</address>
<ref type="url">http://www.cnrs.fr/</ref>
</desc>
</org>
</tutelle>
</tutelles>
</hal:affiliation>
<country>France</country>
<placeName><settlement type="city">Nancy</settlement>
<settlement type="city">Metz</settlement>
<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>
</author>
</analytic>
</biblStruct>
</sourceDesc>
</fileDesc>
<profileDesc><textClass><keywords scheme="mix" xml:lang="en"><term>Active learning</term>
<term>Classification</term>
<term>Data stream</term>
<term>Label noise</term>
<term>Novelty detection</term>
</keywords>
<keywords scheme="mix" xml:lang="fr"><term>Apprentissage actif</term>
<term>Détection de nouveautés</term>
<term>Erreurs d'étiquetage</term>
<term>Flux de données</term>
</keywords>
<keywords scheme="Wicri" type="topic" xml:lang="fr"><term>Classification</term>
</keywords>
</textClass>
</profileDesc>
</teiHeader>
<front><div type="abstract" xml:lang="en">This thesis focuses on machine learning for data classification. To reduce the labelling cost, active learning allows to query the class label of only some important instances from a human labeller. We propose a new uncertainty measure that characterizes the importance of data and improves the performance of active learning compared to the existing uncertainty measures. This measure determines the smallest instance weight to associate with new data, so that the classifier changes its prediction concerning this data. We then consider a setting where the data arrives continuously from an infinite length stream. We propose an adaptive uncertainty threshold that is suitable for active learning in the streaming setting and achieves a compromise between the number of classification errors and the number of required labels. The existing stream-based active learning methods are initialized with some labelled instances that cover all possible classes. However, in many applications, the evolving nature of the stream implies that new classes can appear at any time. We propose an effective method of active detection of novel classes in a multi-class data stream. This method incrementally maintains a feature space area which is covered by the known classes, and detects those instances that are self-similar and external to that area as novel classes. Finally, it is often difficult to get a completely reliable labelling because the human labeller is subject to labelling errors that reduce the performance of the learned classifier. This problem was solved by introducing a measure that reflects the degree of disagreement between the manually given class and the predicted class, and a new informativeness measure that expresses the necessity for a mislabelled instance to be re-labeled by an alternative labeller.</div>
</front>
</TEI>
<hal api="V3"><titleStmt><title xml:lang="en">Classification and active learning from evolving data streams in the presence of uncertain labeling knowledge</title>
<title xml:lang="fr">Classification et apprentissage actif à partir d'un flux de données évolutif en présence d'étiquetage incertain</title>
<author role="aut"><persName><forename type="first">Mohamed-Rafik</forename>
<surname>Bouguelia</surname>
</persName>
<email>mohamed.bouguelia@loria.fr</email>
<ptr type="url" target="http://www.loria.fr/~mbouguel/"></ptr>
<idno type="halauthor">1281537</idno>
<affiliation ref="#struct-206042"></affiliation>
</author>
<editor role="depositor"><persName><forename>Mohamed-Rafik</forename>
<surname>Bouguelia</surname>
</persName>
<email>mohamed.bouguelia@loria.fr</email>
</editor>
</titleStmt>
<editionStmt><edition n="v1" type="current"><date type="whenSubmitted">2016-01-29 19:40:37</date>
<date type="whenModified">2016-02-03 01:03:46</date>
<date type="whenReleased">2016-02-01 08:46:15</date>
<date type="whenProduced">2015-03-25</date>
<date type="whenEndEmbargoed">2016-01-29</date>
<ref type="file" target="https://hal.inria.fr/tel-01262775/document"><date notBefore="2016-01-29"></date>
</ref>
<ref type="file" n="1" target="https://hal.inria.fr/tel-01262775/file/DDOC_T_2015_0034_BOUGUELIA.pdf"><date notBefore="2016-01-29"></date>
</ref>
</edition>
<respStmt><resp>contributor</resp>
<name key="181103"><persName><forename>Mohamed-Rafik</forename>
<surname>Bouguelia</surname>
</persName>
<email>mohamed.bouguelia@loria.fr</email>
</name>
</respStmt>
</editionStmt>
<publicationStmt><distributor>CCSD</distributor>
<idno type="halId">tel-01262775</idno>
<idno type="halUri">https://hal.inria.fr/tel-01262775</idno>
<idno type="halBibtex">bouguelia:tel-01262775</idno>
<idno type="halRefHtml">Intelligence artificielle [cs.AI]. Université de Lorraine, 2015. Français</idno>
<idno type="halRef">Intelligence artificielle [cs.AI]. Université de Lorraine, 2015. Français</idno>
</publicationStmt>
<seriesStmt><idno type="stamp" n="CNRS">CNRS - Centre national de la recherche scientifique</idno>
<idno type="stamp" n="LORIA-TALC" p="LORIA">Traitement automatique des langues et des connaissances</idno>
<idno type="stamp" n="INRIA">INRIA - Institut National de Recherche en Informatique et en Automatique</idno>
<idno type="stamp" n="UNIV-LORRAINE">Université de Lorraine</idno>
<idno type="stamp" n="LORIA2">Publications du LORIA</idno>
<idno type="stamp" n="LORIA">LORIA - Laboratoire Lorrain de Recherche en Informatique et ses Applications</idno>
</seriesStmt>
<notesStmt></notesStmt>
<sourceDesc><biblStruct><analytic><title xml:lang="en">Classification and active learning from evolving data streams in the presence of uncertain labeling knowledge</title>
<title xml:lang="fr">Classification et apprentissage actif à partir d'un flux de données évolutif en présence d'étiquetage incertain</title>
<author role="aut"><persName><forename type="first">Mohamed-Rafik</forename>
<surname>Bouguelia</surname>
</persName>
<email>mohamed.bouguelia@loria.fr</email>
<ptr type="url" target="http://www.loria.fr/~mbouguel/"></ptr>
<idno type="halAuthorId">1281537</idno>
<affiliation ref="#struct-206042"></affiliation>
</author>
</analytic>
<monogr><imprint><date type="dateDefended">2015-03-25</date>
</imprint>
<authority type="institution">Université de Lorraine</authority>
<authority type="school">IAEM (Informatique, Automatique, Électronique - Électrotechnique, Mathématiques)</authority>
<authority type="supervisor">Abdel Belaid</authority>
</monogr>
</biblStruct>
</sourceDesc>
<profileDesc><langUsage><language ident="fr">French</language>
</langUsage>
<textClass><keywords scheme="author"><term xml:lang="en">Classification</term>
<term xml:lang="en">Active learning</term>
<term xml:lang="en">Data stream</term>
<term xml:lang="en">Novelty detection</term>
<term xml:lang="en">Label noise</term>
<term xml:lang="fr">Détection de nouveautés</term>
<term xml:lang="fr">Apprentissage actif</term>
<term xml:lang="fr">Flux de données</term>
<term xml:lang="fr">Erreurs d'étiquetage</term>
</keywords>
<classCode scheme="halDomain" n="info.info-ai">Computer Science [cs]/Artificial Intelligence [cs.AI]</classCode>
<classCode scheme="halTypology" n="THESE">Theses</classCode>
</textClass>
<abstract xml:lang="en">This thesis focuses on machine learning for data classification. To reduce the labelling cost, active learning allows to query the class label of only some important instances from a human labeller. We propose a new uncertainty measure that characterizes the importance of data and improves the performance of active learning compared to the existing uncertainty measures. This measure determines the smallest instance weight to associate with new data, so that the classifier changes its prediction concerning this data. We then consider a setting where the data arrives continuously from an infinite length stream. We propose an adaptive uncertainty threshold that is suitable for active learning in the streaming setting and achieves a compromise between the number of classification errors and the number of required labels. The existing stream-based active learning methods are initialized with some labelled instances that cover all possible classes. However, in many applications, the evolving nature of the stream implies that new classes can appear at any time. We propose an effective method of active detection of novel classes in a multi-class data stream. This method incrementally maintains a feature space area which is covered by the known classes, and detects those instances that are self-similar and external to that area as novel classes. Finally, it is often difficult to get a completely reliable labelling because the human labeller is subject to labelling errors that reduce the performance of the learned classifier. This problem was solved by introducing a measure that reflects the degree of disagreement between the manually given class and the predicted class, and a new informativeness measure that expresses the necessity for a mislabelled instance to be re-labeled by an alternative labeller.</abstract>
<abstract xml:lang="fr">Cette thèse traite de l'apprentissage automatique pour la classification de données. Afin de réduire le coût de l'étiquetage, l'apprentissage actif permet de formuler des requêtes pour demander à un opérateur d'étiqueter seulement quelques données choisies selon un critère d'importance. Nous proposons une nouvelle mesure d'incertitude qui permet de caractériser l'importance des données et qui améliore les performances de l'apprentissage actif par rapport aux mesures existantes. Cette mesure détermine le plus petit poids nécessaire à associer à une nouvelle donnée pour que le classifieur change sa prédiction concernant cette donnée. Nous intégrons ensuite le fait que les données à traiter arrivent en continu dans un flux de longueur infinie. Nous proposons alors un seuil d'incertitude adaptatif qui convient pour un apprentissage actif à partir d'un flux de données et qui réalise un compromis entre le nombre d'erreurs de classification et le nombre d'étiquettes de classes demandées. Les méthodes existantes d'apprentissage actif à partir de flux de données, sont initialisées avec quelques données étiquetées qui couvrent toutes les classes possibles. Cependant, dans de nombreuses applications, la nature évolutive du flux fait que de nouvelles classes peuvent apparaître à tout moment. Nous proposons une méthode efficace de détection active de nouvelles classes dans un flux de données multi-classes. Cette méthode détermine de façon incrémentale une zone couverte par les classes connues, et détecte les données qui sont extérieures à cette zone et proches entre elles, comme étant de nouvelles classes. Enfin, il est souvent difficile d'obtenir un étiquetage totalement fiable car l'opérateur humain est sujet à des erreurs d'étiquetage qui réduisent les performances du classifieur appris. Cette problématique a été résolue par l'introduction d'une mesure qui reflète le degré de désaccord entre la classe donnée manuellement et la classe prédite et une nouvelle mesure d'"informativité" permettant d'exprimer la nécessité pour une donnée mal étiquetée d'être ré-étiquetée par un opérateur alternatif.</abstract>
</profileDesc>
</hal>
</record>
Pour manipuler ce document sous Unix (Dilib)
EXPLOR_STEP=$WICRI_ROOT/Wicri/Lorraine/explor/InforLorV4/Data/Hal/Curation
HfdSelect -h $EXPLOR_STEP/biblio.hfd -nk 001474 | SxmlIndent | more
Ou
HfdSelect -h $EXPLOR_AREA/Data/Hal/Curation/biblio.hfd -nk 001474 | SxmlIndent | more
Pour mettre un lien sur cette page dans le réseau Wicri
{{Explor lien |wiki= Wicri/Lorraine |area= InforLorV4 |flux= Hal |étape= Curation |type= RBID |clé= Hal:tel-01262775 |texte= Classification and active learning from evolving data streams in the presence of uncertain labeling knowledge }}
![]() | This area was generated with Dilib version V0.6.33. | ![]() |