Avoiding peaking phenomenon of the quadratic discriminant function
Identifieur interne : 002B20 ( Istex/Corpus ); précédent : 002B19; suivant : 002B21Avoiding peaking phenomenon of the quadratic discriminant function
Auteurs : Fumitaka Kimura ; Kenji Takashina ; Shinji Tsuruoka ; Yasuji MiyakeSource :
- Systems and Computers in Japan [ 0882-1666 ] ; 1987.
Abstract
The performance of the discriminant function is degraded if the parameters used contain estimation errors. A peaking phenomenon is known in which the classification rate reaches the maximum for a certain feature dimension, but decreases with the further increase of the feature dimension. This paper reports the result of experimental examination of the relations of the classification rates of the Euclid distance, the quadratic discriminant function [1] to the feature dimension. It is shown that the modified quadratic discriminant function corresponds to the discriminant function where the maximum‐likelihood estimation for the covariance matrix of the quadratic discriminant function is replaced by a kind of pseudo‐Bayes estimation, and suppresses the peaking phenomenon.
Url:
DOI: 10.1002/scj.4690180905
Links to Exploration step
ISTEX:C984DCCF7438251DC1706FB0D0124C49FEA9EABBLe document en format XML
<record><TEI wicri:istexFullTextTei="biblStruct"><teiHeader><fileDesc><titleStmt><title xml:lang="en">Avoiding peaking phenomenon of the quadratic discriminant function</title><author><name sortKey="Kimura, Fumitaka" sort="Kimura, Fumitaka" uniqKey="Kimura F" first="Fumitaka" last="Kimura">Fumitaka Kimura</name><affiliation><mods:affiliation>Faculty of Engineering, Mie University, Tsu, Japan 514</mods:affiliation></affiliation></author><author><name sortKey="Takashina, Kenji" sort="Takashina, Kenji" uniqKey="Takashina K" first="Kenji" last="Takashina">Kenji Takashina</name><affiliation><mods:affiliation>Faculty of Engineering, Mie University, Tsu, Japan 514</mods:affiliation></affiliation></author><author><name sortKey="Tsuruoka, Shinji" sort="Tsuruoka, Shinji" uniqKey="Tsuruoka S" first="Shinji" last="Tsuruoka">Shinji Tsuruoka</name><affiliation><mods:affiliation>Faculty of Engineering, Mie University, Tsu, Japan 514</mods:affiliation></affiliation></author><author><name sortKey="Miyake, Yasuji" sort="Miyake, Yasuji" uniqKey="Miyake Y" first="Yasuji" last="Miyake">Yasuji Miyake</name><affiliation><mods:affiliation>Faculty of Engineering, Mie University, Tsu, Japan 514</mods:affiliation></affiliation></author></titleStmt><publicationStmt><idno type="wicri:source">ISTEX</idno><idno type="RBID">ISTEX:C984DCCF7438251DC1706FB0D0124C49FEA9EABB</idno><date when="1987" year="1987">1987</date><idno type="doi">10.1002/scj.4690180905</idno><idno type="url">https://api.istex.fr/document/C984DCCF7438251DC1706FB0D0124C49FEA9EABB/fulltext/pdf</idno><idno type="wicri:Area/Istex/Corpus">002B20</idno></publicationStmt><sourceDesc><biblStruct><analytic><title level="a" type="main" xml:lang="en">Avoiding peaking phenomenon of the quadratic discriminant function</title><author><name sortKey="Kimura, Fumitaka" sort="Kimura, Fumitaka" uniqKey="Kimura F" first="Fumitaka" last="Kimura">Fumitaka Kimura</name><affiliation><mods:affiliation>Faculty of Engineering, Mie University, Tsu, Japan 514</mods:affiliation></affiliation></author><author><name sortKey="Takashina, Kenji" sort="Takashina, Kenji" uniqKey="Takashina K" first="Kenji" last="Takashina">Kenji Takashina</name><affiliation><mods:affiliation>Faculty of Engineering, Mie University, Tsu, Japan 514</mods:affiliation></affiliation></author><author><name sortKey="Tsuruoka, Shinji" sort="Tsuruoka, Shinji" uniqKey="Tsuruoka S" first="Shinji" last="Tsuruoka">Shinji Tsuruoka</name><affiliation><mods:affiliation>Faculty of Engineering, Mie University, Tsu, Japan 514</mods:affiliation></affiliation></author><author><name sortKey="Miyake, Yasuji" sort="Miyake, Yasuji" uniqKey="Miyake Y" first="Yasuji" last="Miyake">Yasuji Miyake</name><affiliation><mods:affiliation>Faculty of Engineering, Mie University, Tsu, Japan 514</mods:affiliation></affiliation></author></analytic><monogr></monogr><series><title level="j">Systems and Computers in Japan</title><title level="j" type="abbrev">Syst. Comp. Jpn.</title><idno type="ISSN">0882-1666</idno><idno type="eISSN">1520-684X</idno><imprint><publisher>Wiley Subscription Services, Inc., A Wiley Company</publisher><pubPlace>New York</pubPlace><date type="published" when="1987">1987</date><biblScope unit="volume">18</biblScope><biblScope unit="issue">9</biblScope><biblScope unit="page" from="44">44</biblScope><biblScope unit="page" to="51">51</biblScope></imprint><idno type="ISSN">0882-1666</idno></series><idno type="istex">C984DCCF7438251DC1706FB0D0124C49FEA9EABB</idno><idno type="DOI">10.1002/scj.4690180905</idno><idno type="ArticleID">SCJ4690180905</idno></biblStruct></sourceDesc><seriesStmt><idno type="ISSN">0882-1666</idno></seriesStmt></fileDesc><profileDesc><textClass></textClass><langUsage><language ident="en">en</language></langUsage></profileDesc></teiHeader><front><div type="abstract" xml:lang="en">The performance of the discriminant function is degraded if the parameters used contain estimation errors. A peaking phenomenon is known in which the classification rate reaches the maximum for a certain feature dimension, but decreases with the further increase of the feature dimension. This paper reports the result of experimental examination of the relations of the classification rates of the Euclid distance, the quadratic discriminant function [1] to the feature dimension. It is shown that the modified quadratic discriminant function corresponds to the discriminant function where the maximum‐likelihood estimation for the covariance matrix of the quadratic discriminant function is replaced by a kind of pseudo‐Bayes estimation, and suppresses the peaking phenomenon.</div></front></TEI><istex><corpusName>wiley</corpusName><author><json:item><name>Fumitaka Kimura</name><affiliations><json:string>Faculty of Engineering, Mie University, Tsu, Japan 514</json:string></affiliations></json:item><json:item><name>Kenji Takashina</name><affiliations><json:string>Faculty of Engineering, Mie University, Tsu, Japan 514</json:string></affiliations></json:item><json:item><name>Shinji Tsuruoka</name><affiliations><json:string>Faculty of Engineering, Mie University, Tsu, Japan 514</json:string></affiliations></json:item><json:item><name>Yasuji Miyake</name><affiliations><json:string>Faculty of Engineering, Mie University, Tsu, Japan 514</json:string></affiliations></json:item></author><articleId><json:string>SCJ4690180905</json:string></articleId><language><json:string>eng</json:string></language><abstract>The performance of the discriminant function is degraded if the parameters used contain estimation errors. A peaking phenomenon is known in which the classification rate reaches the maximum for a certain feature dimension, but decreases with the further increase of the feature dimension. This paper reports the result of experimental examination of the relations of the classification rates of the Euclid distance, the quadratic discriminant function [1] to the feature dimension. It is shown that the modified quadratic discriminant function corresponds to the discriminant function where the maximum‐likelihood estimation for the covariance matrix of the quadratic discriminant function is replaced by a kind of pseudo‐Bayes estimation, and suppresses the peaking phenomenon.</abstract><qualityIndicators><score>4.235</score><pdfVersion>1.3</pdfVersion><pdfPageSize>594 x 828 pts</pdfPageSize><refBibsNative>true</refBibsNative><keywordCount>0</keywordCount><abstractCharCount>778</abstractCharCount><pdfWordCount>2903</pdfWordCount><pdfCharCount>17639</pdfCharCount><pdfPageCount>8</pdfPageCount><abstractWordCount>111</abstractWordCount></qualityIndicators><title>Avoiding peaking phenomenon of the quadratic discriminant function</title><genre.original><json:string>article</json:string></genre.original><genre><json:string>article</json:string></genre><host><volume>18</volume><publisherId><json:string>SCJ</json:string></publisherId><pages><total>8</total><last>51</last><first>44</first></pages><issn><json:string>0882-1666</json:string></issn><issue>9</issue><subject><json:item><value>Article</value></json:item></subject><genre><json:string>journal</json:string></genre><language><json:string>unknown</json:string></language><eissn><json:string>1520-684X</json:string></eissn><title>Systems and Computers in Japan</title><doi><json:string>10.1002/(ISSN)1520-684X</json:string></doi></host><publicationDate>1987</publicationDate><copyrightDate>1987</copyrightDate><doi><json:string>10.1002/scj.4690180905</json:string></doi><id>C984DCCF7438251DC1706FB0D0124C49FEA9EABB</id><fulltext><json:item><original>true</original><mimetype>application/pdf</mimetype><extension>pdf</extension><uri>https://api.istex.fr/document/C984DCCF7438251DC1706FB0D0124C49FEA9EABB/fulltext/pdf</uri></json:item><json:item><original>false</original><mimetype>application/zip</mimetype><extension>zip</extension><uri>https://api.istex.fr/document/C984DCCF7438251DC1706FB0D0124C49FEA9EABB/fulltext/zip</uri></json:item><istex:fulltextTEI uri="https://api.istex.fr/document/C984DCCF7438251DC1706FB0D0124C49FEA9EABB/fulltext/tei"><teiHeader><fileDesc><titleStmt><title level="a" type="main" xml:lang="en">Avoiding peaking phenomenon of the quadratic discriminant function</title></titleStmt><publicationStmt><authority>ISTEX</authority><publisher>Wiley Subscription Services, Inc., A Wiley Company</publisher><pubPlace>New York</pubPlace><availability><p>WILEY</p></availability><date>1987</date></publicationStmt><sourceDesc><biblStruct type="inbook"><analytic><title level="a" type="main" xml:lang="en">Avoiding peaking phenomenon of the quadratic discriminant function</title><author><persName><forename type="first">Fumitaka</forename><surname>Kimura</surname></persName><note type="biography">Member</note><affiliation>Member</affiliation><affiliation>Faculty of Engineering, Mie University, Tsu, Japan 514</affiliation></author><author><persName><forename type="first">Kenji</forename><surname>Takashina</surname></persName><note type="biography">Nonmember</note><affiliation>Nonmember</affiliation><affiliation>Faculty of Engineering, Mie University, Tsu, Japan 514</affiliation></author><author><persName><forename type="first">Shinji</forename><surname>Tsuruoka</surname></persName><affiliation>Faculty of Engineering, Mie University, Tsu, Japan 514</affiliation></author><author><persName><forename type="first">Yasuji</forename><surname>Miyake</surname></persName><note type="biography">Members</note><affiliation>Members</affiliation><affiliation>Faculty of Engineering, Mie University, Tsu, Japan 514</affiliation></author></analytic><monogr><title level="j">Systems and Computers in Japan</title><title level="j" type="abbrev">Syst. Comp. Jpn.</title><idno type="pISSN">0882-1666</idno><idno type="eISSN">1520-684X</idno><idno type="DOI">10.1002/(ISSN)1520-684X</idno><imprint><publisher>Wiley Subscription Services, Inc., A Wiley Company</publisher><pubPlace>New York</pubPlace><date type="published" when="1987"></date><biblScope unit="volume">18</biblScope><biblScope unit="issue">9</biblScope><biblScope unit="page" from="44">44</biblScope><biblScope unit="page" to="51">51</biblScope></imprint></monogr><idno type="istex">C984DCCF7438251DC1706FB0D0124C49FEA9EABB</idno><idno type="DOI">10.1002/scj.4690180905</idno><idno type="ArticleID">SCJ4690180905</idno></biblStruct></sourceDesc></fileDesc><profileDesc><creation><date>1987</date></creation><langUsage><language ident="en">en</language></langUsage><abstract xml:lang="en"><p>The performance of the discriminant function is degraded if the parameters used contain estimation errors. A peaking phenomenon is known in which the classification rate reaches the maximum for a certain feature dimension, but decreases with the further increase of the feature dimension. This paper reports the result of experimental examination of the relations of the classification rates of the Euclid distance, the quadratic discriminant function [1] to the feature dimension. It is shown that the modified quadratic discriminant function corresponds to the discriminant function where the maximum‐likelihood estimation for the covariance matrix of the quadratic discriminant function is replaced by a kind of pseudo‐Bayes estimation, and suppresses the peaking phenomenon.</p></abstract><textClass><keywords scheme="Journal Subject"><list><head>article-category</head><item><term>Article</term></item></list></keywords></textClass></profileDesc><revisionDesc><change when="1987">Published</change></revisionDesc></teiHeader></istex:fulltextTEI><json:item><original>false</original><mimetype>text/plain</mimetype><extension>txt</extension><uri>https://api.istex.fr/document/C984DCCF7438251DC1706FB0D0124C49FEA9EABB/fulltext/txt</uri></json:item></fulltext><metadata><istex:metadataXml wicri:clean="Wiley, elements deleted: body"><istex:xmlDeclaration>version="1.0" encoding="UTF-8" standalone="yes"</istex:xmlDeclaration><istex:document><component version="2.0" type="serialArticle" xml:lang="en"><header><publicationMeta level="product"><publisherInfo><publisherName>Wiley Subscription Services, Inc., A Wiley Company</publisherName><publisherLoc>New York</publisherLoc></publisherInfo><doi registered="yes">10.1002/(ISSN)1520-684X</doi><issn type="print">0882-1666</issn><issn type="electronic">1520-684X</issn><idGroup><id type="product" value="SCJ"></id></idGroup><titleGroup><title type="main" xml:lang="en" sort="SYSTEMS AND COMPUTERS IN JAPAN">Systems and Computers in Japan</title><title type="short">Syst. Comp. Jpn.</title></titleGroup></publicationMeta><publicationMeta level="part" position="90"><doi origin="wiley" registered="yes">10.1002/scj.v18:9</doi><numberingGroup><numbering type="journalVolume" number="18">18</numbering><numbering type="journalIssue">9</numbering></numberingGroup><coverDate startDate="1987">1987</coverDate></publicationMeta><publicationMeta level="unit" type="article" position="5" status="forIssue"><doi origin="wiley" registered="yes">10.1002/scj.4690180905</doi><idGroup><id type="unit" value="SCJ4690180905"></id></idGroup><countGroup><count type="pageTotal" number="8"></count></countGroup><titleGroup><title type="articleCategory">Article</title><title type="tocHeading1">Articles</title></titleGroup><copyright ownership="publisher">Copyright © 1987 Wiley Periodicals, Inc., A Wiley Company</copyright><eventGroup><event type="firstOnline" date="2007-03-21"></event><event type="publishedOnlineFinalForm" date="2007-03-21"></event><event type="xmlConverted" agent="Converter:JWSART34_TO_WML3G version:2.3.2 mode:FullText source:HeaderRef result:HeaderRef" date="2010-03-04"></event><event type="xmlConverted" agent="Converter:WILEY_ML3G_TO_WILEY_ML3GV2 version:3.8.8" date="2014-02-08"></event><event type="xmlConverted" agent="Converter:WML3G_To_WML3G version:4.1.7 mode:FullText,remove_FC" date="2014-11-04"></event></eventGroup><numberingGroup><numbering type="pageFirst">44</numbering><numbering type="pageLast">51</numbering></numberingGroup><linkGroup><link type="toTypesetVersion" href="file:SCJ.SCJ4690180905.pdf"></link></linkGroup></publicationMeta><contentMeta><countGroup><count type="figureTotal" number="10"></count><count type="tableTotal" number="1"></count><count type="referenceTotal" number="9"></count></countGroup><titleGroup><title type="main" xml:lang="en">Avoiding peaking phenomenon of the quadratic discriminant function</title><title type="short" xml:lang="en">Avoiding Peaking Phenomenon</title></titleGroup><creators><creator xml:id="au1" creatorRole="author" affiliationRef="#af1"><personName><givenNames>Fumitaka</givenNames><familyName>Kimura</familyName></personName><jobTitle>Member</jobTitle></creator><creator xml:id="au2" creatorRole="author" affiliationRef="#af1"><personName><givenNames>Kenji</givenNames><familyName>Takashina</familyName></personName><jobTitle>Nonmember</jobTitle></creator><creator xml:id="au3" creatorRole="author" affiliationRef="#af1"><personName><givenNames>Shinji</givenNames><familyName>Tsuruoka</familyName></personName></creator><creator xml:id="au4" creatorRole="author" affiliationRef="#af1"><personName><givenNames>Yasuji</givenNames><familyName>Miyake</familyName></personName><jobTitle>Members</jobTitle></creator></creators><affiliationGroup><affiliation xml:id="af1" countryCode="JP" type="organization"><unparsedAffiliation>Faculty of Engineering, Mie University, Tsu, Japan 514</unparsedAffiliation></affiliation></affiliationGroup><abstractGroup><abstract type="main" xml:lang="en"><title type="main">Abstract</title><p>The performance of the discriminant function is degraded if the parameters used contain estimation errors. A peaking phenomenon is known in which the classification rate reaches the maximum for a certain feature dimension, but decreases with the further increase of the feature dimension. This paper reports the result of experimental examination of the relations of the classification rates of the Euclid distance, the quadratic discriminant function [1] to the feature dimension. It is shown that the modified quadratic discriminant function corresponds to the discriminant function where the maximum‐likelihood estimation for the covariance matrix of the quadratic discriminant function is replaced by a kind of pseudo‐Bayes estimation, and suppresses the peaking phenomenon.</p></abstract></abstractGroup></contentMeta></header></component></istex:document></istex:metadataXml><mods version="3.6"><titleInfo lang="en"><title>Avoiding peaking phenomenon of the quadratic discriminant function</title></titleInfo><titleInfo type="abbreviated" lang="en"><title>Avoiding Peaking Phenomenon</title></titleInfo><titleInfo type="alternative" contentType="CDATA" lang="en"><title>Avoiding peaking phenomenon of the quadratic discriminant function</title></titleInfo><name type="personal"><namePart type="given">Fumitaka</namePart><namePart type="family">Kimura</namePart><affiliation>Faculty of Engineering, Mie University, Tsu, Japan 514</affiliation><role><roleTerm type="text">author</roleTerm></role><description>Member</description></name><name type="personal"><namePart type="given">Kenji</namePart><namePart type="family">Takashina</namePart><affiliation>Faculty of Engineering, Mie University, Tsu, Japan 514</affiliation><role><roleTerm type="text">author</roleTerm></role><description>Nonmember</description></name><name type="personal"><namePart type="given">Shinji</namePart><namePart type="family">Tsuruoka</namePart><affiliation>Faculty of Engineering, Mie University, Tsu, Japan 514</affiliation><role><roleTerm type="text">author</roleTerm></role></name><name type="personal"><namePart type="given">Yasuji</namePart><namePart type="family">Miyake</namePart><affiliation>Faculty of Engineering, Mie University, Tsu, Japan 514</affiliation><role><roleTerm type="text">author</roleTerm></role><description>Members</description></name><typeOfResource>text</typeOfResource><genre type="article" displayLabel="article"></genre><originInfo><publisher>Wiley Subscription Services, Inc., A Wiley Company</publisher><place><placeTerm type="text">New York</placeTerm></place><dateIssued encoding="w3cdtf">1987</dateIssued><copyrightDate encoding="w3cdtf">1987</copyrightDate></originInfo><language><languageTerm type="code" authority="rfc3066">en</languageTerm><languageTerm type="code" authority="iso639-2b">eng</languageTerm></language><physicalDescription><internetMediaType>text/html</internetMediaType><extent unit="figures">10</extent><extent unit="tables">1</extent><extent unit="references">9</extent></physicalDescription><abstract lang="en">The performance of the discriminant function is degraded if the parameters used contain estimation errors. A peaking phenomenon is known in which the classification rate reaches the maximum for a certain feature dimension, but decreases with the further increase of the feature dimension. This paper reports the result of experimental examination of the relations of the classification rates of the Euclid distance, the quadratic discriminant function [1] to the feature dimension. It is shown that the modified quadratic discriminant function corresponds to the discriminant function where the maximum‐likelihood estimation for the covariance matrix of the quadratic discriminant function is replaced by a kind of pseudo‐Bayes estimation, and suppresses the peaking phenomenon.</abstract><relatedItem type="host"><titleInfo><title>Systems and Computers in Japan</title></titleInfo><titleInfo type="abbreviated"><title>Syst. Comp. Jpn.</title></titleInfo><genre type="journal">journal</genre><subject><genre>article-category</genre><topic>Article</topic></subject><identifier type="ISSN">0882-1666</identifier><identifier type="eISSN">1520-684X</identifier><identifier type="DOI">10.1002/(ISSN)1520-684X</identifier><identifier type="PublisherID">SCJ</identifier><part><date>1987</date><detail type="volume"><caption>vol.</caption><number>18</number></detail><detail type="issue"><caption>no.</caption><number>9</number></detail><extent unit="pages"><start>44</start><end>51</end><total>8</total></extent></part></relatedItem><identifier type="istex">C984DCCF7438251DC1706FB0D0124C49FEA9EABB</identifier><identifier type="DOI">10.1002/scj.4690180905</identifier><identifier type="ArticleID">SCJ4690180905</identifier><accessCondition type="use and reproduction" contentType="copyright">Copyright © 1987 Wiley Periodicals, Inc., A Wiley Company</accessCondition><recordInfo><recordContentSource>WILEY</recordContentSource><recordOrigin>Wiley Subscription Services, Inc., A Wiley Company</recordOrigin></recordInfo></mods></metadata><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 002B20 | SxmlIndent | more
Ou
HfdSelect -h $EXPLOR_AREA/Data/Istex/Corpus/biblio.hfd -nk 002B20 | 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:C984DCCF7438251DC1706FB0D0124C49FEA9EABB
|texte= Avoiding peaking phenomenon of the quadratic discriminant function
}}
| This area was generated with Dilib version V0.6.32. |  |