Local distances preserving based manifold learning

•Defining a local distances preserving based manifold learning method.•Defining a neighborhood graph matrix by better description of the local distances.•Defining a new cost function for graph matrix to find the embedded data manifold.•Low sensitivity to the initialization of the parameters.•Definin...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Expert systems with applications 2020-01, Vol.139, p.112860, Article 112860
Hauptverfasser: Hajizadeh, Rassoul, Aghagolzadeh, A., Ezoji, M.
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page 112860
container_title Expert systems with applications
container_volume 139
creator Hajizadeh, Rassoul
Aghagolzadeh, A.
Ezoji, M.
description •Defining a local distances preserving based manifold learning method.•Defining a neighborhood graph matrix by better description of the local distances.•Defining a new cost function for graph matrix to find the embedded data manifold.•Low sensitivity to the initialization of the parameters.•Defining a criterion to evaluate the local manifold learning methods. In this paper a local manifold learning method based on the local distances preserving (LDP) is proposed. LDP focuses on extracting and preserving the local distances between the data points. In LDP, the coefficients between each data point and its neighbors are calculated based on the inverse of Euclidean distances between them. Then, a minimization function, which is in accordance with the calculated coefficients, is proposed for calculating the embedded data manifold in the low-dimensional representation space. Matching the way of calculation the coefficients and the proposed minimization function is the main significance of the proposed LDP method. In addition, the proposed LDP method shows less sensitivity to the initial parameters such as the number of neighbors and the value of variance. Also, a stochastic based extension of LDP (SLDP) manifold learning method is proposed. The proposed method is compared with the common manifold learning methods based on the achievable recognition rate and the power of the local distances preserving. The experiments have been done on two different kinds of databases: HODA Persian handwritten character database and ORL face image database. The results demonstrate the suitable performance of LDP and SLDP. Also, the results show the robustness of the proposed method to the number of neighbors and the value of variance parameter. Moreover, the proposed method of calculating the embedded data points in LDP and SLDP has less complexity than the similar local manifold learning methods, Laplacian eigenmaps (LEM) and stochastic LEM (SLEM).
doi_str_mv 10.1016/j.eswa.2019.112860
format Article
fullrecord <record><control><sourceid>proquest_cross</sourceid><recordid>TN_cdi_proquest_journals_2312231914</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><els_id>S0957417419305706</els_id><sourcerecordid>2312231914</sourcerecordid><originalsourceid>FETCH-LOGICAL-c328t-903d5ff81955925e9b15abd7525f47569981acee839a1cf3bf747433214e04f33</originalsourceid><addsrcrecordid>eNp9kE1LxDAQhoMouK7-AU8Fz62ZfDQNeBHxCxa86DmkyURSuu2adFf893apZw_DwPC-M-88hFwDrYBCfdtVmL9txSjoCoA1NT0hK2gUL2ul-SlZUS1VKUCJc3KRc0cpKErVirDN6Gxf-JgnOzjMxS5hxnSIw2fR2oy-2NohhrH3RY82DfP8kpwF22e8-utr8vH0-P7wUm7enl8f7jel46yZSk25lyE0oKXUTKJuQdrWK8lkEErWWjdgHWLDtQUXeBuUUIJzBgKpCJyvyc2yd5fGrz3myXTjPg3zScM4sLk0iFnFFpVLY84Jg9mluLXpxwA1RzamM0c25sjGLGxm091iwjn_IWIy2UWc__cxoZuMH-N_9l8E_Gtf</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2312231914</pqid></control><display><type>article</type><title>Local distances preserving based manifold learning</title><source>Elsevier ScienceDirect Journals Complete</source><creator>Hajizadeh, Rassoul ; Aghagolzadeh, A. ; Ezoji, M.</creator><creatorcontrib>Hajizadeh, Rassoul ; Aghagolzadeh, A. ; Ezoji, M.</creatorcontrib><description>•Defining a local distances preserving based manifold learning method.•Defining a neighborhood graph matrix by better description of the local distances.•Defining a new cost function for graph matrix to find the embedded data manifold.•Low sensitivity to the initialization of the parameters.•Defining a criterion to evaluate the local manifold learning methods. In this paper a local manifold learning method based on the local distances preserving (LDP) is proposed. LDP focuses on extracting and preserving the local distances between the data points. In LDP, the coefficients between each data point and its neighbors are calculated based on the inverse of Euclidean distances between them. Then, a minimization function, which is in accordance with the calculated coefficients, is proposed for calculating the embedded data manifold in the low-dimensional representation space. Matching the way of calculation the coefficients and the proposed minimization function is the main significance of the proposed LDP method. In addition, the proposed LDP method shows less sensitivity to the initial parameters such as the number of neighbors and the value of variance. Also, a stochastic based extension of LDP (SLDP) manifold learning method is proposed. The proposed method is compared with the common manifold learning methods based on the achievable recognition rate and the power of the local distances preserving. The experiments have been done on two different kinds of databases: HODA Persian handwritten character database and ORL face image database. The results demonstrate the suitable performance of LDP and SLDP. Also, the results show the robustness of the proposed method to the number of neighbors and the value of variance parameter. Moreover, the proposed method of calculating the embedded data points in LDP and SLDP has less complexity than the similar local manifold learning methods, Laplacian eigenmaps (LEM) and stochastic LEM (SLEM).</description><identifier>ISSN: 0957-4174</identifier><identifier>EISSN: 1873-6793</identifier><identifier>DOI: 10.1016/j.eswa.2019.112860</identifier><language>eng</language><publisher>New York: Elsevier Ltd</publisher><subject>Coefficients ; Data points ; Dimension reduction ; Euclidean distance ; Handwriting ; Local distance ; Machine learning ; Manifold learning ; Manifolds (mathematics) ; Mathematical analysis ; Optimization ; Parameter sensitivity ; Recognition rate ; Teaching methods</subject><ispartof>Expert systems with applications, 2020-01, Vol.139, p.112860, Article 112860</ispartof><rights>2019</rights><rights>Copyright Elsevier BV Jan 2020</rights><lds50>peer_reviewed</lds50><woscitedreferencessubscribed>false</woscitedreferencessubscribed><citedby>FETCH-LOGICAL-c328t-903d5ff81955925e9b15abd7525f47569981acee839a1cf3bf747433214e04f33</citedby><cites>FETCH-LOGICAL-c328t-903d5ff81955925e9b15abd7525f47569981acee839a1cf3bf747433214e04f33</cites><orcidid>0000-0002-6999-3464 ; 0000-0003-0471-4396</orcidid></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktohtml>$$Uhttps://dx.doi.org/10.1016/j.eswa.2019.112860$$EHTML$$P50$$Gelsevier$$H</linktohtml><link.rule.ids>314,780,784,3550,27924,27925,45995</link.rule.ids></links><search><creatorcontrib>Hajizadeh, Rassoul</creatorcontrib><creatorcontrib>Aghagolzadeh, A.</creatorcontrib><creatorcontrib>Ezoji, M.</creatorcontrib><title>Local distances preserving based manifold learning</title><title>Expert systems with applications</title><description>•Defining a local distances preserving based manifold learning method.•Defining a neighborhood graph matrix by better description of the local distances.•Defining a new cost function for graph matrix to find the embedded data manifold.•Low sensitivity to the initialization of the parameters.•Defining a criterion to evaluate the local manifold learning methods. In this paper a local manifold learning method based on the local distances preserving (LDP) is proposed. LDP focuses on extracting and preserving the local distances between the data points. In LDP, the coefficients between each data point and its neighbors are calculated based on the inverse of Euclidean distances between them. Then, a minimization function, which is in accordance with the calculated coefficients, is proposed for calculating the embedded data manifold in the low-dimensional representation space. Matching the way of calculation the coefficients and the proposed minimization function is the main significance of the proposed LDP method. In addition, the proposed LDP method shows less sensitivity to the initial parameters such as the number of neighbors and the value of variance. Also, a stochastic based extension of LDP (SLDP) manifold learning method is proposed. The proposed method is compared with the common manifold learning methods based on the achievable recognition rate and the power of the local distances preserving. The experiments have been done on two different kinds of databases: HODA Persian handwritten character database and ORL face image database. The results demonstrate the suitable performance of LDP and SLDP. Also, the results show the robustness of the proposed method to the number of neighbors and the value of variance parameter. Moreover, the proposed method of calculating the embedded data points in LDP and SLDP has less complexity than the similar local manifold learning methods, Laplacian eigenmaps (LEM) and stochastic LEM (SLEM).</description><subject>Coefficients</subject><subject>Data points</subject><subject>Dimension reduction</subject><subject>Euclidean distance</subject><subject>Handwriting</subject><subject>Local distance</subject><subject>Machine learning</subject><subject>Manifold learning</subject><subject>Manifolds (mathematics)</subject><subject>Mathematical analysis</subject><subject>Optimization</subject><subject>Parameter sensitivity</subject><subject>Recognition rate</subject><subject>Teaching methods</subject><issn>0957-4174</issn><issn>1873-6793</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2020</creationdate><recordtype>article</recordtype><recordid>eNp9kE1LxDAQhoMouK7-AU8Fz62ZfDQNeBHxCxa86DmkyURSuu2adFf893apZw_DwPC-M-88hFwDrYBCfdtVmL9txSjoCoA1NT0hK2gUL2ul-SlZUS1VKUCJc3KRc0cpKErVirDN6Gxf-JgnOzjMxS5hxnSIw2fR2oy-2NohhrH3RY82DfP8kpwF22e8-utr8vH0-P7wUm7enl8f7jel46yZSk25lyE0oKXUTKJuQdrWK8lkEErWWjdgHWLDtQUXeBuUUIJzBgKpCJyvyc2yd5fGrz3myXTjPg3zScM4sLk0iFnFFpVLY84Jg9mluLXpxwA1RzamM0c25sjGLGxm091iwjn_IWIy2UWc__cxoZuMH-N_9l8E_Gtf</recordid><startdate>202001</startdate><enddate>202001</enddate><creator>Hajizadeh, Rassoul</creator><creator>Aghagolzadeh, A.</creator><creator>Ezoji, M.</creator><general>Elsevier Ltd</general><general>Elsevier BV</general><scope>AAYXX</scope><scope>CITATION</scope><scope>7SC</scope><scope>8FD</scope><scope>JQ2</scope><scope>L7M</scope><scope>L~C</scope><scope>L~D</scope><orcidid>https://orcid.org/0000-0002-6999-3464</orcidid><orcidid>https://orcid.org/0000-0003-0471-4396</orcidid></search><sort><creationdate>202001</creationdate><title>Local distances preserving based manifold learning</title><author>Hajizadeh, Rassoul ; Aghagolzadeh, A. ; Ezoji, M.</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c328t-903d5ff81955925e9b15abd7525f47569981acee839a1cf3bf747433214e04f33</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2020</creationdate><topic>Coefficients</topic><topic>Data points</topic><topic>Dimension reduction</topic><topic>Euclidean distance</topic><topic>Handwriting</topic><topic>Local distance</topic><topic>Machine learning</topic><topic>Manifold learning</topic><topic>Manifolds (mathematics)</topic><topic>Mathematical analysis</topic><topic>Optimization</topic><topic>Parameter sensitivity</topic><topic>Recognition rate</topic><topic>Teaching methods</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Hajizadeh, Rassoul</creatorcontrib><creatorcontrib>Aghagolzadeh, A.</creatorcontrib><creatorcontrib>Ezoji, M.</creatorcontrib><collection>CrossRef</collection><collection>Computer and Information Systems Abstracts</collection><collection>Technology Research Database</collection><collection>ProQuest Computer Science Collection</collection><collection>Advanced Technologies Database with Aerospace</collection><collection>Computer and Information Systems Abstracts – Academic</collection><collection>Computer and Information Systems Abstracts Professional</collection><jtitle>Expert systems with applications</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Hajizadeh, Rassoul</au><au>Aghagolzadeh, A.</au><au>Ezoji, M.</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Local distances preserving based manifold learning</atitle><jtitle>Expert systems with applications</jtitle><date>2020-01</date><risdate>2020</risdate><volume>139</volume><spage>112860</spage><pages>112860-</pages><artnum>112860</artnum><issn>0957-4174</issn><eissn>1873-6793</eissn><abstract>•Defining a local distances preserving based manifold learning method.•Defining a neighborhood graph matrix by better description of the local distances.•Defining a new cost function for graph matrix to find the embedded data manifold.•Low sensitivity to the initialization of the parameters.•Defining a criterion to evaluate the local manifold learning methods. In this paper a local manifold learning method based on the local distances preserving (LDP) is proposed. LDP focuses on extracting and preserving the local distances between the data points. In LDP, the coefficients between each data point and its neighbors are calculated based on the inverse of Euclidean distances between them. Then, a minimization function, which is in accordance with the calculated coefficients, is proposed for calculating the embedded data manifold in the low-dimensional representation space. Matching the way of calculation the coefficients and the proposed minimization function is the main significance of the proposed LDP method. In addition, the proposed LDP method shows less sensitivity to the initial parameters such as the number of neighbors and the value of variance. Also, a stochastic based extension of LDP (SLDP) manifold learning method is proposed. The proposed method is compared with the common manifold learning methods based on the achievable recognition rate and the power of the local distances preserving. The experiments have been done on two different kinds of databases: HODA Persian handwritten character database and ORL face image database. The results demonstrate the suitable performance of LDP and SLDP. Also, the results show the robustness of the proposed method to the number of neighbors and the value of variance parameter. Moreover, the proposed method of calculating the embedded data points in LDP and SLDP has less complexity than the similar local manifold learning methods, Laplacian eigenmaps (LEM) and stochastic LEM (SLEM).</abstract><cop>New York</cop><pub>Elsevier Ltd</pub><doi>10.1016/j.eswa.2019.112860</doi><orcidid>https://orcid.org/0000-0002-6999-3464</orcidid><orcidid>https://orcid.org/0000-0003-0471-4396</orcidid></addata></record>
fulltext fulltext
identifier ISSN: 0957-4174
ispartof Expert systems with applications, 2020-01, Vol.139, p.112860, Article 112860
issn 0957-4174
1873-6793
language eng
recordid cdi_proquest_journals_2312231914
source Elsevier ScienceDirect Journals Complete
subjects Coefficients
Data points
Dimension reduction
Euclidean distance
Handwriting
Local distance
Machine learning
Manifold learning
Manifolds (mathematics)
Mathematical analysis
Optimization
Parameter sensitivity
Recognition rate
Teaching methods
title Local distances preserving based manifold learning
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-03T18%3A13%3A20IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest_cross&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Local%20distances%20preserving%20based%20manifold%20learning&rft.jtitle=Expert%20systems%20with%20applications&rft.au=Hajizadeh,%20Rassoul&rft.date=2020-01&rft.volume=139&rft.spage=112860&rft.pages=112860-&rft.artnum=112860&rft.issn=0957-4174&rft.eissn=1873-6793&rft_id=info:doi/10.1016/j.eswa.2019.112860&rft_dat=%3Cproquest_cross%3E2312231914%3C/proquest_cross%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=2312231914&rft_id=info:pmid/&rft_els_id=S0957417419305706&rfr_iscdi=true