Loading…

Hierarchical One-Class Model With Subnetwork for Representation Learning and Outlier Detection

The multilayer one-class classification (OCC) frameworks have gained great traction in research on anomaly and outlier detection. However, most multilayer OCC algorithms suffer from loosely connected feature coding, affecting the ability of generated latent space to properly generate a highly discri...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transactions on cybernetics 2023-10, Vol.53 (10), p.6303-6316
Main Authors: Zhang, Wandong, Wu, Q. M. Jonathan, Zhao, W. G. Will, Deng, Haojin, Yang, Yimin
Format: Article
Language:English
Subjects:
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
cited_by cdi_FETCH-LOGICAL-c349t-9c7bc408d416c00995f945693c336853742e1b06573f44856d8598fdf182c3383
cites cdi_FETCH-LOGICAL-c349t-9c7bc408d416c00995f945693c336853742e1b06573f44856d8598fdf182c3383
container_end_page 6316
container_issue 10
container_start_page 6303
container_title IEEE transactions on cybernetics
container_volume 53
creator Zhang, Wandong
Wu, Q. M. Jonathan
Zhao, W. G. Will
Deng, Haojin
Yang, Yimin
description The multilayer one-class classification (OCC) frameworks have gained great traction in research on anomaly and outlier detection. However, most multilayer OCC algorithms suffer from loosely connected feature coding, affecting the ability of generated latent space to properly generate a highly discriminative representation between object classes. To alleviate this deficiency, two novel OCC frameworks, namely: 1) OCC structure using the subnetwork neural network (OC-SNN) and 2) maximum correntropy-based OC-SNN (MCOC-SNN), are proposed in this article. The novelties of this article are as follows: 1) the subnetwork is used to build the discriminative latent space; 2) the proposed models are one-step learning networks, instead of stacking feature learning blocks and final classification layer to recognize the input pattern; 3) unlike existing works which utilize mean square error (MSE) to learn low-dimensional features, the MCOC-SNN uses maximum correntropy criterion (MCC) for discriminative feature encoding; and 4) a brand-new OCC dataset, called CO-Mask, is built for this research. Experimental results on the visual classification domain with a varying number of training samples from 6131 to 513 061 demonstrate that the proposed OC-SNN and MCOC-SNN achieve superior performance compared to the existing multilayer OCC models. For reproducibility, the source codes are available at https://github.com/W1AE/OCC .
doi_str_mv 10.1109/TCYB.2022.3166349
format article
fullrecord <record><control><sourceid>proquest_pubme</sourceid><recordid>TN_cdi_proquest_journals_2865091990</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><ieee_id>9765789</ieee_id><sourcerecordid>2658227520</sourcerecordid><originalsourceid>FETCH-LOGICAL-c349t-9c7bc408d416c00995f945693c336853742e1b06573f44856d8598fdf182c3383</originalsourceid><addsrcrecordid>eNpdkE1LJDEQhoOsqKg_QBaWwF689JjvTo674yeMDKyKeDFk0tVruz3pMelG9t-bZsY5mEtC1VMvlQehE0omlBJzdj99-j1hhLEJp0pxYXbQAaNKF4yV8tv2rcp9dJzSK8lH55LRe2ifS6GVVOIAPV83EF30L413LZ4HKKatSwnfdhW0-LHpX_DdsAjQv3fxH667iP_AKkKC0Lu-6QKegYuhCX-xCxWeD32b8_A59ODH9hHarV2b4HhzH6KHy4v76XUxm1_dTH_NCp8X7wvjy4UXRFeCKk-IMbI2QirDPedKS14KBnRBlCx5LYSWqtLS6LqqqWYZ0fwQna5zV7F7GyD1dtkkD23rAnRDskxJPXphJKM_v6Cv3RBD3s6yLIUYasxI0TXlY5dShNquYrN08b-lxI7-7ejfjv7txn-e-bFJHhZLqLYTn7Yz8H0NNACwbZsy_0sb_gEgkYcz</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2865091990</pqid></control><display><type>article</type><title>Hierarchical One-Class Model With Subnetwork for Representation Learning and Outlier Detection</title><source>IEEE Xplore (Online service)</source><creator>Zhang, Wandong ; Wu, Q. M. Jonathan ; Zhao, W. G. Will ; Deng, Haojin ; Yang, Yimin</creator><creatorcontrib>Zhang, Wandong ; Wu, Q. M. Jonathan ; Zhao, W. G. Will ; Deng, Haojin ; Yang, Yimin</creatorcontrib><description>The multilayer one-class classification (OCC) frameworks have gained great traction in research on anomaly and outlier detection. However, most multilayer OCC algorithms suffer from loosely connected feature coding, affecting the ability of generated latent space to properly generate a highly discriminative representation between object classes. To alleviate this deficiency, two novel OCC frameworks, namely: 1) OCC structure using the subnetwork neural network (OC-SNN) and 2) maximum correntropy-based OC-SNN (MCOC-SNN), are proposed in this article. The novelties of this article are as follows: 1) the subnetwork is used to build the discriminative latent space; 2) the proposed models are one-step learning networks, instead of stacking feature learning blocks and final classification layer to recognize the input pattern; 3) unlike existing works which utilize mean square error (MSE) to learn low-dimensional features, the MCOC-SNN uses maximum correntropy criterion (MCC) for discriminative feature encoding; and 4) a brand-new OCC dataset, called CO-Mask, is built for this research. Experimental results on the visual classification domain with a varying number of training samples from 6131 to 513 061 demonstrate that the proposed OC-SNN and MCOC-SNN achieve superior performance compared to the existing multilayer OCC models. For reproducibility, the source codes are available at https://github.com/W1AE/OCC .</description><identifier>ISSN: 2168-2267</identifier><identifier>ISSN: 2168-2275</identifier><identifier>EISSN: 2168-2275</identifier><identifier>DOI: 10.1109/TCYB.2022.3166349</identifier><identifier>PMID: 35486564</identifier><identifier>CODEN: ITCEB8</identifier><language>eng</language><publisher>United States: IEEE</publisher><subject>Algorithms ; Anomaly detection ; Classification ; Data analysis ; Data models ; Encoding ; Hierarchical network ; Machine learning ; Moore--Penrose inverse (MPI) ; Multilayers ; Neural networks ; Nonhomogeneous media ; one-class classification (OCC) ; Outliers (statistics) ; Representation learning ; Representations ; Task analysis ; Training</subject><ispartof>IEEE transactions on cybernetics, 2023-10, Vol.53 (10), p.6303-6316</ispartof><rights>Copyright The Institute of Electrical and Electronics Engineers, Inc. (IEEE) 2023</rights><lds50>peer_reviewed</lds50><woscitedreferencessubscribed>false</woscitedreferencessubscribed><citedby>FETCH-LOGICAL-c349t-9c7bc408d416c00995f945693c336853742e1b06573f44856d8598fdf182c3383</citedby><cites>FETCH-LOGICAL-c349t-9c7bc408d416c00995f945693c336853742e1b06573f44856d8598fdf182c3383</cites><orcidid>0000-0002-5208-7975 ; 0000-0002-5083-5052 ; 0000-0003-2534-6879 ; 0000-0002-1131-2056</orcidid></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktohtml>$$Uhttps://ieeexplore.ieee.org/document/9765789$$EHTML$$P50$$Gieee$$H</linktohtml><link.rule.ids>314,780,784,27924,27925,54796</link.rule.ids><backlink>$$Uhttps://www.ncbi.nlm.nih.gov/pubmed/35486564$$D View this record in MEDLINE/PubMed$$Hfree_for_read</backlink></links><search><creatorcontrib>Zhang, Wandong</creatorcontrib><creatorcontrib>Wu, Q. M. Jonathan</creatorcontrib><creatorcontrib>Zhao, W. G. Will</creatorcontrib><creatorcontrib>Deng, Haojin</creatorcontrib><creatorcontrib>Yang, Yimin</creatorcontrib><title>Hierarchical One-Class Model With Subnetwork for Representation Learning and Outlier Detection</title><title>IEEE transactions on cybernetics</title><addtitle>TCYB</addtitle><addtitle>IEEE Trans Cybern</addtitle><description>The multilayer one-class classification (OCC) frameworks have gained great traction in research on anomaly and outlier detection. However, most multilayer OCC algorithms suffer from loosely connected feature coding, affecting the ability of generated latent space to properly generate a highly discriminative representation between object classes. To alleviate this deficiency, two novel OCC frameworks, namely: 1) OCC structure using the subnetwork neural network (OC-SNN) and 2) maximum correntropy-based OC-SNN (MCOC-SNN), are proposed in this article. The novelties of this article are as follows: 1) the subnetwork is used to build the discriminative latent space; 2) the proposed models are one-step learning networks, instead of stacking feature learning blocks and final classification layer to recognize the input pattern; 3) unlike existing works which utilize mean square error (MSE) to learn low-dimensional features, the MCOC-SNN uses maximum correntropy criterion (MCC) for discriminative feature encoding; and 4) a brand-new OCC dataset, called CO-Mask, is built for this research. Experimental results on the visual classification domain with a varying number of training samples from 6131 to 513 061 demonstrate that the proposed OC-SNN and MCOC-SNN achieve superior performance compared to the existing multilayer OCC models. For reproducibility, the source codes are available at https://github.com/W1AE/OCC .</description><subject>Algorithms</subject><subject>Anomaly detection</subject><subject>Classification</subject><subject>Data analysis</subject><subject>Data models</subject><subject>Encoding</subject><subject>Hierarchical network</subject><subject>Machine learning</subject><subject>Moore--Penrose inverse (MPI)</subject><subject>Multilayers</subject><subject>Neural networks</subject><subject>Nonhomogeneous media</subject><subject>one-class classification (OCC)</subject><subject>Outliers (statistics)</subject><subject>Representation learning</subject><subject>Representations</subject><subject>Task analysis</subject><subject>Training</subject><issn>2168-2267</issn><issn>2168-2275</issn><issn>2168-2275</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2023</creationdate><recordtype>article</recordtype><recordid>eNpdkE1LJDEQhoOsqKg_QBaWwF689JjvTo674yeMDKyKeDFk0tVruz3pMelG9t-bZsY5mEtC1VMvlQehE0omlBJzdj99-j1hhLEJp0pxYXbQAaNKF4yV8tv2rcp9dJzSK8lH55LRe2ifS6GVVOIAPV83EF30L413LZ4HKKatSwnfdhW0-LHpX_DdsAjQv3fxH667iP_AKkKC0Lu-6QKegYuhCX-xCxWeD32b8_A59ODH9hHarV2b4HhzH6KHy4v76XUxm1_dTH_NCp8X7wvjy4UXRFeCKk-IMbI2QirDPedKS14KBnRBlCx5LYSWqtLS6LqqqWYZ0fwQna5zV7F7GyD1dtkkD23rAnRDskxJPXphJKM_v6Cv3RBD3s6yLIUYasxI0TXlY5dShNquYrN08b-lxI7-7ejfjv7txn-e-bFJHhZLqLYTn7Yz8H0NNACwbZsy_0sb_gEgkYcz</recordid><startdate>20231001</startdate><enddate>20231001</enddate><creator>Zhang, Wandong</creator><creator>Wu, Q. M. Jonathan</creator><creator>Zhao, W. G. Will</creator><creator>Deng, Haojin</creator><creator>Yang, Yimin</creator><general>IEEE</general><general>The Institute of Electrical and Electronics Engineers, Inc. (IEEE)</general><scope>97E</scope><scope>RIA</scope><scope>RIE</scope><scope>NPM</scope><scope>AAYXX</scope><scope>CITATION</scope><scope>7SC</scope><scope>7SP</scope><scope>7TB</scope><scope>8FD</scope><scope>F28</scope><scope>FR3</scope><scope>H8D</scope><scope>JQ2</scope><scope>L7M</scope><scope>L~C</scope><scope>L~D</scope><scope>7X8</scope><orcidid>https://orcid.org/0000-0002-5208-7975</orcidid><orcidid>https://orcid.org/0000-0002-5083-5052</orcidid><orcidid>https://orcid.org/0000-0003-2534-6879</orcidid><orcidid>https://orcid.org/0000-0002-1131-2056</orcidid></search><sort><creationdate>20231001</creationdate><title>Hierarchical One-Class Model With Subnetwork for Representation Learning and Outlier Detection</title><author>Zhang, Wandong ; Wu, Q. M. Jonathan ; Zhao, W. G. Will ; Deng, Haojin ; Yang, Yimin</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c349t-9c7bc408d416c00995f945693c336853742e1b06573f44856d8598fdf182c3383</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2023</creationdate><topic>Algorithms</topic><topic>Anomaly detection</topic><topic>Classification</topic><topic>Data analysis</topic><topic>Data models</topic><topic>Encoding</topic><topic>Hierarchical network</topic><topic>Machine learning</topic><topic>Moore--Penrose inverse (MPI)</topic><topic>Multilayers</topic><topic>Neural networks</topic><topic>Nonhomogeneous media</topic><topic>one-class classification (OCC)</topic><topic>Outliers (statistics)</topic><topic>Representation learning</topic><topic>Representations</topic><topic>Task analysis</topic><topic>Training</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Zhang, Wandong</creatorcontrib><creatorcontrib>Wu, Q. M. Jonathan</creatorcontrib><creatorcontrib>Zhao, W. G. Will</creatorcontrib><creatorcontrib>Deng, Haojin</creatorcontrib><creatorcontrib>Yang, Yimin</creatorcontrib><collection>IEEE All-Society Periodicals Package (ASPP) 2005-present</collection><collection>IEEE All-Society Periodicals Package (ASPP) 1998–Present</collection><collection>IEEE Xplore (Online service)</collection><collection>PubMed</collection><collection>CrossRef</collection><collection>Computer and Information Systems Abstracts</collection><collection>Electronics &amp; Communications Abstracts</collection><collection>Mechanical &amp; Transportation Engineering Abstracts</collection><collection>Technology Research Database</collection><collection>ANTE: Abstracts in New Technology &amp; Engineering</collection><collection>Engineering Research Database</collection><collection>Aerospace 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><collection>MEDLINE - Academic</collection><jtitle>IEEE transactions on cybernetics</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Zhang, Wandong</au><au>Wu, Q. M. Jonathan</au><au>Zhao, W. G. Will</au><au>Deng, Haojin</au><au>Yang, Yimin</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Hierarchical One-Class Model With Subnetwork for Representation Learning and Outlier Detection</atitle><jtitle>IEEE transactions on cybernetics</jtitle><stitle>TCYB</stitle><addtitle>IEEE Trans Cybern</addtitle><date>2023-10-01</date><risdate>2023</risdate><volume>53</volume><issue>10</issue><spage>6303</spage><epage>6316</epage><pages>6303-6316</pages><issn>2168-2267</issn><issn>2168-2275</issn><eissn>2168-2275</eissn><coden>ITCEB8</coden><abstract>The multilayer one-class classification (OCC) frameworks have gained great traction in research on anomaly and outlier detection. However, most multilayer OCC algorithms suffer from loosely connected feature coding, affecting the ability of generated latent space to properly generate a highly discriminative representation between object classes. To alleviate this deficiency, two novel OCC frameworks, namely: 1) OCC structure using the subnetwork neural network (OC-SNN) and 2) maximum correntropy-based OC-SNN (MCOC-SNN), are proposed in this article. The novelties of this article are as follows: 1) the subnetwork is used to build the discriminative latent space; 2) the proposed models are one-step learning networks, instead of stacking feature learning blocks and final classification layer to recognize the input pattern; 3) unlike existing works which utilize mean square error (MSE) to learn low-dimensional features, the MCOC-SNN uses maximum correntropy criterion (MCC) for discriminative feature encoding; and 4) a brand-new OCC dataset, called CO-Mask, is built for this research. Experimental results on the visual classification domain with a varying number of training samples from 6131 to 513 061 demonstrate that the proposed OC-SNN and MCOC-SNN achieve superior performance compared to the existing multilayer OCC models. For reproducibility, the source codes are available at https://github.com/W1AE/OCC .</abstract><cop>United States</cop><pub>IEEE</pub><pmid>35486564</pmid><doi>10.1109/TCYB.2022.3166349</doi><tpages>14</tpages><orcidid>https://orcid.org/0000-0002-5208-7975</orcidid><orcidid>https://orcid.org/0000-0002-5083-5052</orcidid><orcidid>https://orcid.org/0000-0003-2534-6879</orcidid><orcidid>https://orcid.org/0000-0002-1131-2056</orcidid></addata></record>
fulltext fulltext
identifier ISSN: 2168-2267
ispartof IEEE transactions on cybernetics, 2023-10, Vol.53 (10), p.6303-6316
issn 2168-2267
2168-2275
2168-2275
language eng
recordid cdi_proquest_journals_2865091990
source IEEE Xplore (Online service)
subjects Algorithms
Anomaly detection
Classification
Data analysis
Data models
Encoding
Hierarchical network
Machine learning
Moore--Penrose inverse (MPI)
Multilayers
Neural networks
Nonhomogeneous media
one-class classification (OCC)
Outliers (statistics)
Representation learning
Representations
Task analysis
Training
title Hierarchical One-Class Model With Subnetwork for Representation Learning and Outlier Detection
url http://sfxeu10.hosted.exlibrisgroup.com/loughborough?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-24T20%3A08%3A57IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest_pubme&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Hierarchical%20One-Class%20Model%20With%20Subnetwork%20for%20Representation%20Learning%20and%20Outlier%20Detection&rft.jtitle=IEEE%20transactions%20on%20cybernetics&rft.au=Zhang,%20Wandong&rft.date=2023-10-01&rft.volume=53&rft.issue=10&rft.spage=6303&rft.epage=6316&rft.pages=6303-6316&rft.issn=2168-2267&rft.eissn=2168-2275&rft.coden=ITCEB8&rft_id=info:doi/10.1109/TCYB.2022.3166349&rft_dat=%3Cproquest_pubme%3E2658227520%3C/proquest_pubme%3E%3Cgrp_id%3Ecdi_FETCH-LOGICAL-c349t-9c7bc408d416c00995f945693c336853742e1b06573f44856d8598fdf182c3383%3C/grp_id%3E%3Coa%3E%3C/oa%3E%3Curl%3E%3C/url%3E&rft_id=info:oai/&rft_pqid=2865091990&rft_id=info:pmid/35486564&rft_ieee_id=9765789&rfr_iscdi=true