Pattern classification by a condensed neural network
Identifieur interne : 001939 ( Istex/Corpus ); précédent : 001938; suivant : 001940Pattern classification by a condensed neural network
Auteurs : A. Mitiche ; M. LebidoffSource :
- Neural Networks [ 0893-6080 ] ; 2001.
English descriptors
- KwdEn :
Abstract
Neural networks have come to the fore as potent pattern classifiers. More amenable to parallel computation, they are much faster than the nearest neighbor classifier (NN), which, however, has distinctly outperformed them in several applications. The purpose of this study is to investigate a condensed neural network that combines the classification speed of neural networks and the low error rate of the nearest neighbor classifier. This condensed network is a fast, accurate classifier of simple architecture and function: it consists of a set of generalized perceptrons that draw maximal hyperspherical boundaries centered on patterns of memory units, each circumscribing reference patterns of a single category. The generalized perceptrons carry out classification, assisted by sporadic nearest neighbor matching to patterns of a small reference set. We compare the condensed network to a high performance neural network pattern classifier (Kohonen) and to NN in experiments on hand-printed character recognition.
Url:
DOI: 10.1016/S0893-6080(01)00035-1
Links to Exploration step
ISTEX:E3542D2652F2E9515D988C14C3035FA0F24A6746Le document en format XML
<record><TEI wicri:istexFullTextTei="biblStruct"><teiHeader><fileDesc><titleStmt><title xml:lang="en">Pattern classification by a condensed neural network</title><author><name sortKey="Mitiche, A" sort="Mitiche, A" uniqKey="Mitiche A" first="A" last="Mitiche">A. Mitiche</name><affiliation><mods:affiliation>INRS-Télécommunications, Place Bonaventure, 900, de la Gauchetière Ouest, Niveau C, Montréal, Québec, Canada, H5A 1C6</mods:affiliation></affiliation><affiliation><mods:affiliation>E-mail: mitiche@inrs-telecom.uquebec.ca</mods:affiliation></affiliation></author><author><name sortKey="Lebidoff, M" sort="Lebidoff, M" uniqKey="Lebidoff M" first="M" last="Lebidoff">M. Lebidoff</name><affiliation><mods:affiliation>INRS-Télécommunications, Place Bonaventure, 900, de la Gauchetière Ouest, Niveau C, Montréal, Québec, Canada, H5A 1C6</mods:affiliation></affiliation></author></titleStmt><publicationStmt><idno type="wicri:source">ISTEX</idno><idno type="RBID">ISTEX:E3542D2652F2E9515D988C14C3035FA0F24A6746</idno><date when="2001" year="2001">2001</date><idno type="doi">10.1016/S0893-6080(01)00035-1</idno><idno type="url">https://api.istex.fr/document/E3542D2652F2E9515D988C14C3035FA0F24A6746/fulltext/pdf</idno><idno type="wicri:Area/Istex/Corpus">001939</idno></publicationStmt><sourceDesc><biblStruct><analytic><title level="a" type="main" xml:lang="en">Pattern classification by a condensed neural network</title><author><name sortKey="Mitiche, A" sort="Mitiche, A" uniqKey="Mitiche A" first="A" last="Mitiche">A. Mitiche</name><affiliation><mods:affiliation>INRS-Télécommunications, Place Bonaventure, 900, de la Gauchetière Ouest, Niveau C, Montréal, Québec, Canada, H5A 1C6</mods:affiliation></affiliation><affiliation><mods:affiliation>E-mail: mitiche@inrs-telecom.uquebec.ca</mods:affiliation></affiliation></author><author><name sortKey="Lebidoff, M" sort="Lebidoff, M" uniqKey="Lebidoff M" first="M" last="Lebidoff">M. Lebidoff</name><affiliation><mods:affiliation>INRS-Télécommunications, Place Bonaventure, 900, de la Gauchetière Ouest, Niveau C, Montréal, Québec, Canada, H5A 1C6</mods:affiliation></affiliation></author></analytic><monogr></monogr><series><title level="j">Neural Networks</title><title level="j" type="abbrev">NN</title><idno type="ISSN">0893-6080</idno><imprint><publisher>ELSEVIER</publisher><date type="published" when="2001">2001</date><biblScope unit="volume">14</biblScope><biblScope unit="issue">4–5</biblScope><biblScope unit="page" from="575">575</biblScope><biblScope unit="page" to="580">580</biblScope></imprint><idno type="ISSN">0893-6080</idno></series><idno type="istex">E3542D2652F2E9515D988C14C3035FA0F24A6746</idno><idno type="DOI">10.1016/S0893-6080(01)00035-1</idno><idno type="PII">S0893-6080(01)00035-1</idno></biblStruct></sourceDesc><seriesStmt><idno type="ISSN">0893-6080</idno></seriesStmt></fileDesc><profileDesc><textClass><keywords scheme="KwdEn" xml:lang="en"><term>Condensed neural network</term><term>Hand-printed character recognition</term><term>Kohonen network</term><term>Nearest neighbor classifier</term><term>Neural networks</term><term>OCR</term><term>Pattern classification</term></keywords></textClass><langUsage><language ident="en">en</language></langUsage></profileDesc></teiHeader><front><div type="abstract" xml:lang="en">Neural networks have come to the fore as potent pattern classifiers. More amenable to parallel computation, they are much faster than the nearest neighbor classifier (NN), which, however, has distinctly outperformed them in several applications. The purpose of this study is to investigate a condensed neural network that combines the classification speed of neural networks and the low error rate of the nearest neighbor classifier. This condensed network is a fast, accurate classifier of simple architecture and function: it consists of a set of generalized perceptrons that draw maximal hyperspherical boundaries centered on patterns of memory units, each circumscribing reference patterns of a single category. The generalized perceptrons carry out classification, assisted by sporadic nearest neighbor matching to patterns of a small reference set. We compare the condensed network to a high performance neural network pattern classifier (Kohonen) and to NN in experiments on hand-printed character recognition.</div></front></TEI><istex><corpusName>elsevier</corpusName><author><json:item><name>A Mitiche</name><affiliations><json:string>INRS-Télécommunications, Place Bonaventure, 900, de la Gauchetière Ouest, Niveau C, Montréal, Québec, Canada, H5A 1C6</json:string><json:string>E-mail: mitiche@inrs-telecom.uquebec.ca</json:string></affiliations></json:item><json:item><name>M Lebidoff</name><affiliations><json:string>INRS-Télécommunications, Place Bonaventure, 900, de la Gauchetière Ouest, Niveau C, Montréal, Québec, Canada, H5A 1C6</json:string></affiliations></json:item></author><subject><json:item><lang><json:string>eng</json:string></lang><value>Neural networks</value></json:item><json:item><lang><json:string>eng</json:string></lang><value>Condensed neural network</value></json:item><json:item><lang><json:string>eng</json:string></lang><value>Kohonen network</value></json:item><json:item><lang><json:string>eng</json:string></lang><value>Nearest neighbor classifier</value></json:item><json:item><lang><json:string>eng</json:string></lang><value>Pattern classification</value></json:item><json:item><lang><json:string>eng</json:string></lang><value>Hand-printed character recognition</value></json:item><json:item><lang><json:string>eng</json:string></lang><value>OCR</value></json:item></subject><language><json:string>eng</json:string></language><abstract>Neural networks have come to the fore as potent pattern classifiers. More amenable to parallel computation, they are much faster than the nearest neighbor classifier (NN), which, however, has distinctly outperformed them in several applications. The purpose of this study is to investigate a condensed neural network that combines the classification speed of neural networks and the low error rate of the nearest neighbor classifier. This condensed network is a fast, accurate classifier of simple architecture and function: it consists of a set of generalized perceptrons that draw maximal hyperspherical boundaries centered on patterns of memory units, each circumscribing reference patterns of a single category. The generalized perceptrons carry out classification, assisted by sporadic nearest neighbor matching to patterns of a small reference set. We compare the condensed network to a high performance neural network pattern classifier (Kohonen) and to NN in experiments on hand-printed character recognition.</abstract><qualityIndicators><score>5.435</score><pdfVersion>1.2</pdfVersion><pdfPageSize>595 x 794 pts</pdfPageSize><refBibsNative>true</refBibsNative><keywordCount>7</keywordCount><abstractCharCount>1017</abstractCharCount><pdfWordCount>3671</pdfWordCount><pdfCharCount>21493</pdfCharCount><pdfPageCount>6</pdfPageCount><abstractWordCount>147</abstractWordCount></qualityIndicators><title>Pattern classification by a condensed neural network</title><pii><json:string>S0893-6080(01)00035-1</json:string></pii><genre><json:string>research-article</json:string></genre><host><volume>14</volume><pii><json:string>S0893-6080(00)X0053-6</json:string></pii><pages><last>580</last><first>575</first></pages><issn><json:string>0893-6080</json:string></issn><issue>4–5</issue><genre><json:string>Journal</json:string></genre><language><json:string>unknown</json:string></language><title>Neural Networks</title><publicationDate>2001</publicationDate></host><categories><wos><json:string>COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE</json:string><json:string>NEUROSCIENCES</json:string></wos></categories><publicationDate>2001</publicationDate><copyrightDate>2001</copyrightDate><doi><json:string>10.1016/S0893-6080(01)00035-1</json:string></doi><id>E3542D2652F2E9515D988C14C3035FA0F24A6746</id><fulltext><json:item><original>true</original><mimetype>application/pdf</mimetype><extension>pdf</extension><uri>https://api.istex.fr/document/E3542D2652F2E9515D988C14C3035FA0F24A6746/fulltext/pdf</uri></json:item><json:item><original>true</original><mimetype>text/plain</mimetype><extension>txt</extension><uri>https://api.istex.fr/document/E3542D2652F2E9515D988C14C3035FA0F24A6746/fulltext/txt</uri></json:item><json:item><original>false</original><mimetype>application/zip</mimetype><extension>zip</extension><uri>https://api.istex.fr/document/E3542D2652F2E9515D988C14C3035FA0F24A6746/fulltext/zip</uri></json:item><istex:fulltextTEI uri="https://api.istex.fr/document/E3542D2652F2E9515D988C14C3035FA0F24A6746/fulltext/tei"><teiHeader><fileDesc><titleStmt><title level="a" type="main" xml:lang="en">Pattern classification by a condensed neural network</title></titleStmt><publicationStmt><authority>ISTEX</authority><publisher>ELSEVIER</publisher><availability><p>ELSEVIER</p></availability><date>2001</date></publicationStmt><notesStmt><note type="content">Section title: Contributed article</note><note type="content">Fig. 1: Perceptron Pj implementing discriminant function gj. The cj's and rj are learned from training.</note><note type="content">Fig. 2: A is the set of all nodes of memory if the perceptrons’ outputs are all zero, i.e. if oj=0 for all j=1, …, m. Otherwise, A is the subset of nodes of memory that includes center Cj if oj=1. NN is the rule that assigns to X the class k of the closest pattern in A.</note><note type="content">Table 1:</note><note type="content">Table 2:</note><note type="content">Table 3:</note></notesStmt><sourceDesc><biblStruct type="inbook"><analytic><title level="a" type="main" xml:lang="en">Pattern classification by a condensed neural network</title><author><persName><forename type="first">A</forename><surname>Mitiche</surname></persName><email>mitiche@inrs-telecom.uquebec.ca</email><note type="correspondence"><p>Corresponding author</p></note><affiliation>INRS-Télécommunications, Place Bonaventure, 900, de la Gauchetière Ouest, Niveau C, Montréal, Québec, Canada, H5A 1C6</affiliation></author><author><persName><forename type="first">M</forename><surname>Lebidoff</surname></persName><affiliation>INRS-Télécommunications, Place Bonaventure, 900, de la Gauchetière Ouest, Niveau C, Montréal, Québec, Canada, H5A 1C6</affiliation></author></analytic><monogr><title level="j">Neural Networks</title><title level="j" type="abbrev">NN</title><idno type="pISSN">0893-6080</idno><idno type="PII">S0893-6080(00)X0053-6</idno><imprint><publisher>ELSEVIER</publisher><date type="published" when="2001"></date><biblScope unit="volume">14</biblScope><biblScope unit="issue">4–5</biblScope><biblScope unit="page" from="575">575</biblScope><biblScope unit="page" to="580">580</biblScope></imprint></monogr><idno type="istex">E3542D2652F2E9515D988C14C3035FA0F24A6746</idno><idno type="DOI">10.1016/S0893-6080(01)00035-1</idno><idno type="PII">S0893-6080(01)00035-1</idno></biblStruct></sourceDesc></fileDesc><profileDesc><creation><date>2001</date></creation><langUsage><language ident="en">en</language></langUsage><abstract xml:lang="en"><p>Neural networks have come to the fore as potent pattern classifiers. More amenable to parallel computation, they are much faster than the nearest neighbor classifier (NN), which, however, has distinctly outperformed them in several applications. The purpose of this study is to investigate a condensed neural network that combines the classification speed of neural networks and the low error rate of the nearest neighbor classifier. This condensed network is a fast, accurate classifier of simple architecture and function: it consists of a set of generalized perceptrons that draw maximal hyperspherical boundaries centered on patterns of memory units, each circumscribing reference patterns of a single category. The generalized perceptrons carry out classification, assisted by sporadic nearest neighbor matching to patterns of a small reference set. We compare the condensed network to a high performance neural network pattern classifier (Kohonen) and to NN in experiments on hand-printed character recognition.</p></abstract><textClass xml:lang="en"><keywords scheme="keyword"><list><head>Keywords</head><item><term>Neural networks</term></item><item><term>Condensed neural network</term></item><item><term>Kohonen network</term></item><item><term>Nearest neighbor classifier</term></item><item><term>Pattern classification</term></item><item><term>Hand-printed character recognition</term></item><item><term>OCR</term></item></list></keywords></textClass></profileDesc><revisionDesc><change when="2001-02-22">Registration</change><change when="2001-02-22">Modified</change><change when="2001">Published</change></revisionDesc></teiHeader></istex:fulltextTEI></fulltext><metadata><istex:metadataXml wicri:clean="Elsevier, elements deleted: ce:floats; body; tail"><istex:xmlDeclaration>version="1.0" encoding="utf-8"</istex:xmlDeclaration><istex:docType PUBLIC="-//ES//DTD journal article DTD version 4.5.2//EN//XML" URI="art452.dtd" name="istex:docType"><istex:entity SYSTEM="gr1" NDATA="IMAGE" name="gr1"></istex:entity><istex:entity SYSTEM="gr2" NDATA="IMAGE" name="gr2"></istex:entity></istex:docType><istex:document><converted-article version="4.5.2" docsubtype="fla" xml:lang="en"><item-info><jid>NN</jid><aid>1498</aid><ce:pii>S0893-6080(01)00035-1</ce:pii><ce:doi>10.1016/S0893-6080(01)00035-1</ce:doi><ce:copyright type="full-transfer" year="2001">Elsevier Science Ltd</ce:copyright></item-info><head><ce:dochead><ce:textfn>Contributed article</ce:textfn></ce:dochead><ce:title>Pattern classification by a condensed neural network</ce:title><ce:author-group><ce:author><ce:given-name>A</ce:given-name><ce:surname>Mitiche</ce:surname><ce:cross-ref refid="CORR1">*</ce:cross-ref><ce:e-address>mitiche@inrs-telecom.uquebec.ca</ce:e-address></ce:author><ce:author><ce:given-name>M</ce:given-name><ce:surname>Lebidoff</ce:surname></ce:author><ce:affiliation><ce:textfn>INRS-Télécommunications, Place Bonaventure, 900, de la Gauchetière Ouest, Niveau C, Montréal, Québec, Canada, H5A 1C6</ce:textfn></ce:affiliation><ce:correspondence id="CORR1"><ce:label>*</ce:label><ce:text>Corresponding author</ce:text></ce:correspondence></ce:author-group><ce:date-received day="15" month="6" year="2000"></ce:date-received><ce:date-revised day="22" month="2" year="2001"></ce:date-revised><ce:date-accepted day="22" month="2" year="2001"></ce:date-accepted><ce:abstract><ce:section-title>Abstract</ce:section-title><ce:abstract-sec><ce:simple-para>Neural networks have come to the fore as potent pattern classifiers. More amenable to parallel computation, they are much faster than the nearest neighbor classifier (NN), which, however, has distinctly outperformed them in several applications. The purpose of this study is to investigate a condensed neural network that combines the classification speed of neural networks and the low error rate of the nearest neighbor classifier. This condensed network is a fast, accurate classifier of simple architecture and function: it consists of a set of generalized perceptrons that draw maximal hyperspherical boundaries centered on patterns of memory units, each circumscribing reference patterns of a single category. The generalized perceptrons carry out classification, assisted by sporadic nearest neighbor matching to patterns of a small reference set. We compare the condensed network to a high performance neural network pattern classifier (Kohonen) and to NN in experiments on hand-printed character recognition.</ce:simple-para></ce:abstract-sec></ce:abstract><ce:keywords class="keyword" xml:lang="en"><ce:section-title>Keywords</ce:section-title><ce:keyword><ce:text>Neural networks</ce:text></ce:keyword><ce:keyword><ce:text>Condensed neural network</ce:text></ce:keyword><ce:keyword><ce:text>Kohonen network</ce:text></ce:keyword><ce:keyword><ce:text>Nearest neighbor classifier</ce:text></ce:keyword><ce:keyword><ce:text>Pattern classification</ce:text></ce:keyword><ce:keyword><ce:text>Hand-printed character recognition</ce:text></ce:keyword><ce:keyword><ce:text>OCR</ce:text></ce:keyword></ce:keywords></head></converted-article></istex:document></istex:metadataXml><mods version="3.6"><titleInfo lang="en"><title>Pattern classification by a condensed neural network</title></titleInfo><titleInfo type="alternative" lang="en" contentType="CDATA"><title>Pattern classification by a condensed neural network</title></titleInfo><name type="personal"><namePart type="given">A</namePart><namePart type="family">Mitiche</namePart><affiliation>INRS-Télécommunications, Place Bonaventure, 900, de la Gauchetière Ouest, Niveau C, Montréal, Québec, Canada, H5A 1C6</affiliation><affiliation>E-mail: mitiche@inrs-telecom.uquebec.ca</affiliation><description>Corresponding author</description><role><roleTerm type="text">author</roleTerm></role></name><name type="personal"><namePart type="given">M</namePart><namePart type="family">Lebidoff</namePart><affiliation>INRS-Télécommunications, Place Bonaventure, 900, de la Gauchetière Ouest, Niveau C, Montréal, Québec, Canada, H5A 1C6</affiliation><role><roleTerm type="text">author</roleTerm></role></name><typeOfResource>text</typeOfResource><genre type="research-article" displayLabel="Full-length article"></genre><originInfo><publisher>ELSEVIER</publisher><dateIssued encoding="w3cdtf">2001</dateIssued><dateValid encoding="w3cdtf">2001-02-22</dateValid><dateModified encoding="w3cdtf">2001-02-22</dateModified><copyrightDate encoding="w3cdtf">2001</copyrightDate></originInfo><language><languageTerm type="code" authority="iso639-2b">eng</languageTerm><languageTerm type="code" authority="rfc3066">en</languageTerm></language><physicalDescription><internetMediaType>text/html</internetMediaType></physicalDescription><abstract lang="en">Neural networks have come to the fore as potent pattern classifiers. More amenable to parallel computation, they are much faster than the nearest neighbor classifier (NN), which, however, has distinctly outperformed them in several applications. The purpose of this study is to investigate a condensed neural network that combines the classification speed of neural networks and the low error rate of the nearest neighbor classifier. This condensed network is a fast, accurate classifier of simple architecture and function: it consists of a set of generalized perceptrons that draw maximal hyperspherical boundaries centered on patterns of memory units, each circumscribing reference patterns of a single category. The generalized perceptrons carry out classification, assisted by sporadic nearest neighbor matching to patterns of a small reference set. We compare the condensed network to a high performance neural network pattern classifier (Kohonen) and to NN in experiments on hand-printed character recognition.</abstract><note type="content">Section title: Contributed article</note><note type="content">Fig. 1: Perceptron Pj implementing discriminant function gj. The cj's and rj are learned from training.</note><note type="content">Fig. 2: A is the set of all nodes of memory if the perceptrons’ outputs are all zero, i.e. if oj=0 for all j=1, …, m. Otherwise, A is the subset of nodes of memory that includes center Cj if oj=1. NN is the rule that assigns to X the class k of the closest pattern in A.</note><note type="content">Table 1: </note><note type="content">Table 2: </note><note type="content">Table 3: </note><subject lang="en"><genre>Keywords</genre><topic>Neural networks</topic><topic>Condensed neural network</topic><topic>Kohonen network</topic><topic>Nearest neighbor classifier</topic><topic>Pattern classification</topic><topic>Hand-printed character recognition</topic><topic>OCR</topic></subject><relatedItem type="host"><titleInfo><title>Neural Networks</title></titleInfo><titleInfo type="abbreviated"><title>NN</title></titleInfo><genre type="Journal">journal</genre><originInfo><dateIssued encoding="w3cdtf">200105</dateIssued></originInfo><identifier type="ISSN">0893-6080</identifier><identifier type="PII">S0893-6080(00)X0053-6</identifier><part><date>200105</date><detail type="volume"><number>14</number><caption>vol.</caption></detail><detail type="issue"><number>4–5</number><caption>no.</caption></detail><extent unit="issue pages"><start>381</start><end>586</end></extent><extent unit="pages"><start>575</start><end>580</end></extent></part></relatedItem><identifier type="istex">E3542D2652F2E9515D988C14C3035FA0F24A6746</identifier><identifier type="DOI">10.1016/S0893-6080(01)00035-1</identifier><identifier type="PII">S0893-6080(01)00035-1</identifier><accessCondition type="use and reproduction" contentType="">© 2001Elsevier Science Ltd</accessCondition><recordInfo><recordContentSource>ELSEVIER</recordContentSource><recordOrigin>Elsevier Science Ltd, ©2001</recordOrigin></recordInfo></mods></metadata><enrichments><istex:catWosTEI uri="https://api.istex.fr/document/E3542D2652F2E9515D988C14C3035FA0F24A6746/enrichments/catWos"><teiHeader><profileDesc><textClass><classCode scheme="WOS">COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE</classCode><classCode scheme="WOS">NEUROSCIENCES</classCode></textClass></profileDesc></teiHeader></istex:catWosTEI></enrichments><serie></serie></istex></record>Pour manipuler ce document sous Unix (Dilib)
EXPLOR_STEP=$WICRI_ROOT/Ticri/CIDE/explor/OcrV1/Data/Istex/Corpus
HfdSelect -h $EXPLOR_STEP/biblio.hfd -nk 001939 | SxmlIndent | more
Ou
HfdSelect -h $EXPLOR_AREA/Data/Istex/Corpus/biblio.hfd -nk 001939 | SxmlIndent | more
Pour mettre un lien sur cette page dans le réseau Wicri
{{Explor lien
|wiki= Ticri/CIDE
|area= OcrV1
|flux= Istex
|étape= Corpus
|type= RBID
|clé= ISTEX:E3542D2652F2E9515D988C14C3035FA0F24A6746
|texte= Pattern classification by a condensed neural network
}}
| This area was generated with Dilib version V0.6.32. |  |