Loading…
Problematic Projection to the In-Sample Subspace for a Kernelized Anomaly Detector
We examine the properties and performance of kernelized anomaly detectors, with an emphasis on the Mahalanobis-distance-based kernel RX (KRX) algorithm. Although the detector generally performs well for high-bandwidth Gaussian kernels, it exhibits problematic (in some cases, catastrophic) performanc...
Saved in:
| Published in: | IEEE geoscience and remote sensing letters 2016-04, Vol.13 (4), p.485-489 |
|---|---|
| Main Authors: | , |
| 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-c353t-3e6617fee7336a105147bbde3cdbba645e6d29a6c9265e75eaccd6d212c46a6f3 |
|---|---|
| cites | cdi_FETCH-LOGICAL-c353t-3e6617fee7336a105147bbde3cdbba645e6d29a6c9265e75eaccd6d212c46a6f3 |
| container_end_page | 489 |
| container_issue | 4 |
| container_start_page | 485 |
| container_title | IEEE geoscience and remote sensing letters |
| container_volume | 13 |
| creator | Theiler, James Grosklos, Guen |
| description | We examine the properties and performance of kernelized anomaly detectors, with an emphasis on the Mahalanobis-distance-based kernel RX (KRX) algorithm. Although the detector generally performs well for high-bandwidth Gaussian kernels, it exhibits problematic (in some cases, catastrophic) performance for distances that are large compared to the bandwidth. By comparing KRX to two other anomaly detectors, we can trace the problem to a projection in feature space, which arises when a pseudoinverse is used on the covariance matrix in that feature space. We show that a regularized variant of KRX overcomes this difficulty and achieves superior performance over a wide range of bandwidths. |
| doi_str_mv | 10.1109/LGRS.2016.2516985 |
| format | article |
| fullrecord | <record><control><sourceid>proquest_ieee_</sourceid><recordid>TN_cdi_ieee_primary_7426793</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><ieee_id>7426793</ieee_id><sourcerecordid>4046245741</sourcerecordid><originalsourceid>FETCH-LOGICAL-c353t-3e6617fee7336a105147bbde3cdbba645e6d29a6c9265e75eaccd6d212c46a6f3</originalsourceid><addsrcrecordid>eNpdkctOwzAQRSMEEqXwAYiNBRs2KR47tpNlVZ6iEqgFiZ3lOBM1VRIXO13A1-OqiAWruRqdOw_dJDkHOgGgxc38YbGcMApywgTIIhcHyQiEyFMqFBzudCZSUeQfx8lJCGtKWZbnapQsXr0rW-zM0FgS9Rrt0LieDI4MKyRPfbo03aZFstyWYWMsktp5Ysgz-h7b5hsrMu1dZ9ovcotDNDt_mhzVpg149lvHyfv93dvsMZ2_PDzNpvPUcsGHlKOUoGpExbk0QAVkqiwr5LYqSyMzgbJihZG2YFKgEmisrWILmM2kkTUfJ5f7uS4MjQ62ietX1vV9vEIDy5SURYSu99DGu88thkF3TbDYtqZHtw0acpCUSyiyiF79Q9du6_v4ggaVK8pyARAp2FPWuxA81nrjm874Lw1U76LQuyj0Lgr9G0X0XOw9DSL-8SpjUhWc_wCcNoQ_</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>1787028511</pqid></control><display><type>article</type><title>Problematic Projection to the In-Sample Subspace for a Kernelized Anomaly Detector</title><source>IEEE Electronic Library (IEL) Journals</source><creator>Theiler, James ; Grosklos, Guen</creator><creatorcontrib>Theiler, James ; Grosklos, Guen ; Los Alamos National Lab. (LANL), Los Alamos, NM (United States)</creatorcontrib><description>We examine the properties and performance of kernelized anomaly detectors, with an emphasis on the Mahalanobis-distance-based kernel RX (KRX) algorithm. Although the detector generally performs well for high-bandwidth Gaussian kernels, it exhibits problematic (in some cases, catastrophic) performance for distances that are large compared to the bandwidth. By comparing KRX to two other anomaly detectors, we can trace the problem to a projection in feature space, which arises when a pseudoinverse is used on the covariance matrix in that feature space. We show that a regularized variant of KRX overcomes this difficulty and achieves superior performance over a wide range of bandwidths.</description><identifier>ISSN: 1545-598X</identifier><identifier>EISSN: 1558-0571</identifier><identifier>DOI: 10.1109/LGRS.2016.2516985</identifier><identifier>CODEN: IGRSBY</identifier><language>eng</language><publisher>Piscataway: IEEE</publisher><subject>Adaptive signal detection ; Algorithms ; Anomalies ; Anomaly detection ; Bandwidth ; Covariance matrices ; data models ; Detectors ; Gaussian ; Gaussian distribution ; GEOSCIENCES ; Hyperspectral imaging ; kernel density estimation ; kernel-RX ; Kernels ; mahalanobis distance ; MATHEMATICS AND COMPUTING ; multidimensional signal processing ; pattern recognition ; Projection ; Remote sensing ; singular value decomposition ; spectral analysis ; Unsupervised learning</subject><ispartof>IEEE geoscience and remote sensing letters, 2016-04, Vol.13 (4), p.485-489</ispartof><rights>Copyright The Institute of Electrical and Electronics Engineers, Inc. (IEEE) 2016</rights><lds50>peer_reviewed</lds50><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed><citedby>FETCH-LOGICAL-c353t-3e6617fee7336a105147bbde3cdbba645e6d29a6c9265e75eaccd6d212c46a6f3</citedby><cites>FETCH-LOGICAL-c353t-3e6617fee7336a105147bbde3cdbba645e6d29a6c9265e75eaccd6d212c46a6f3</cites></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktohtml>$$Uhttps://ieeexplore.ieee.org/document/7426793$$EHTML$$P50$$Gieee$$Hfree_for_read</linktohtml><link.rule.ids>230,314,780,784,885,27924,27925,54796</link.rule.ids><backlink>$$Uhttps://www.osti.gov/servlets/purl/1247669$$D View this record in Osti.gov$$Hfree_for_read</backlink></links><search><creatorcontrib>Theiler, James</creatorcontrib><creatorcontrib>Grosklos, Guen</creatorcontrib><creatorcontrib>Los Alamos National Lab. (LANL), Los Alamos, NM (United States)</creatorcontrib><title>Problematic Projection to the In-Sample Subspace for a Kernelized Anomaly Detector</title><title>IEEE geoscience and remote sensing letters</title><addtitle>LGRS</addtitle><description>We examine the properties and performance of kernelized anomaly detectors, with an emphasis on the Mahalanobis-distance-based kernel RX (KRX) algorithm. Although the detector generally performs well for high-bandwidth Gaussian kernels, it exhibits problematic (in some cases, catastrophic) performance for distances that are large compared to the bandwidth. By comparing KRX to two other anomaly detectors, we can trace the problem to a projection in feature space, which arises when a pseudoinverse is used on the covariance matrix in that feature space. We show that a regularized variant of KRX overcomes this difficulty and achieves superior performance over a wide range of bandwidths.</description><subject>Adaptive signal detection</subject><subject>Algorithms</subject><subject>Anomalies</subject><subject>Anomaly detection</subject><subject>Bandwidth</subject><subject>Covariance matrices</subject><subject>data models</subject><subject>Detectors</subject><subject>Gaussian</subject><subject>Gaussian distribution</subject><subject>GEOSCIENCES</subject><subject>Hyperspectral imaging</subject><subject>kernel density estimation</subject><subject>kernel-RX</subject><subject>Kernels</subject><subject>mahalanobis distance</subject><subject>MATHEMATICS AND COMPUTING</subject><subject>multidimensional signal processing</subject><subject>pattern recognition</subject><subject>Projection</subject><subject>Remote sensing</subject><subject>singular value decomposition</subject><subject>spectral analysis</subject><subject>Unsupervised learning</subject><issn>1545-598X</issn><issn>1558-0571</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2016</creationdate><recordtype>article</recordtype><sourceid>ESBDL</sourceid><recordid>eNpdkctOwzAQRSMEEqXwAYiNBRs2KR47tpNlVZ6iEqgFiZ3lOBM1VRIXO13A1-OqiAWruRqdOw_dJDkHOgGgxc38YbGcMApywgTIIhcHyQiEyFMqFBzudCZSUeQfx8lJCGtKWZbnapQsXr0rW-zM0FgS9Rrt0LieDI4MKyRPfbo03aZFstyWYWMsktp5Ysgz-h7b5hsrMu1dZ9ovcotDNDt_mhzVpg149lvHyfv93dvsMZ2_PDzNpvPUcsGHlKOUoGpExbk0QAVkqiwr5LYqSyMzgbJihZG2YFKgEmisrWILmM2kkTUfJ5f7uS4MjQ62ietX1vV9vEIDy5SURYSu99DGu88thkF3TbDYtqZHtw0acpCUSyiyiF79Q9du6_v4ggaVK8pyARAp2FPWuxA81nrjm874Lw1U76LQuyj0Lgr9G0X0XOw9DSL-8SpjUhWc_wCcNoQ_</recordid><startdate>20160401</startdate><enddate>20160401</enddate><creator>Theiler, James</creator><creator>Grosklos, Guen</creator><general>IEEE</general><general>The Institute of Electrical and Electronics Engineers, Inc. (IEEE)</general><scope>97E</scope><scope>ESBDL</scope><scope>RIA</scope><scope>RIE</scope><scope>AAYXX</scope><scope>CITATION</scope><scope>7SC</scope><scope>7SP</scope><scope>7TG</scope><scope>7UA</scope><scope>8FD</scope><scope>C1K</scope><scope>F1W</scope><scope>FR3</scope><scope>H8D</scope><scope>H96</scope><scope>JQ2</scope><scope>KL.</scope><scope>KR7</scope><scope>L.G</scope><scope>L7M</scope><scope>L~C</scope><scope>L~D</scope><scope>F28</scope><scope>OIOZB</scope><scope>OTOTI</scope></search><sort><creationdate>20160401</creationdate><title>Problematic Projection to the In-Sample Subspace for a Kernelized Anomaly Detector</title><author>Theiler, James ; Grosklos, Guen</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c353t-3e6617fee7336a105147bbde3cdbba645e6d29a6c9265e75eaccd6d212c46a6f3</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2016</creationdate><topic>Adaptive signal detection</topic><topic>Algorithms</topic><topic>Anomalies</topic><topic>Anomaly detection</topic><topic>Bandwidth</topic><topic>Covariance matrices</topic><topic>data models</topic><topic>Detectors</topic><topic>Gaussian</topic><topic>Gaussian distribution</topic><topic>GEOSCIENCES</topic><topic>Hyperspectral imaging</topic><topic>kernel density estimation</topic><topic>kernel-RX</topic><topic>Kernels</topic><topic>mahalanobis distance</topic><topic>MATHEMATICS AND COMPUTING</topic><topic>multidimensional signal processing</topic><topic>pattern recognition</topic><topic>Projection</topic><topic>Remote sensing</topic><topic>singular value decomposition</topic><topic>spectral analysis</topic><topic>Unsupervised learning</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Theiler, James</creatorcontrib><creatorcontrib>Grosklos, Guen</creatorcontrib><creatorcontrib>Los Alamos National Lab. (LANL), Los Alamos, NM (United States)</creatorcontrib><collection>IEEE All-Society Periodicals Package (ASPP) 2005-present</collection><collection>IEEE Open Access Journals</collection><collection>IEEE All-Society Periodicals Package (ASPP) 1998-Present</collection><collection>IEEE/IET Electronic Library</collection><collection>CrossRef</collection><collection>Computer and Information Systems Abstracts</collection><collection>Electronics & Communications Abstracts</collection><collection>Meteorological & Geoastrophysical Abstracts</collection><collection>Water Resources Abstracts</collection><collection>Technology Research Database</collection><collection>Environmental Sciences and Pollution Management</collection><collection>ASFA: Aquatic Sciences and Fisheries Abstracts</collection><collection>Engineering Research Database</collection><collection>Aerospace Database</collection><collection>Aquatic Science & Fisheries Abstracts (ASFA) 2: Ocean Technology, Policy & Non-Living Resources</collection><collection>ProQuest Computer Science Collection</collection><collection>Meteorological & Geoastrophysical Abstracts - Academic</collection><collection>Civil Engineering Abstracts</collection><collection>Aquatic Science & Fisheries Abstracts (ASFA) Professional</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>ANTE: Abstracts in New Technology & Engineering</collection><collection>OSTI.GOV - Hybrid</collection><collection>OSTI.GOV</collection><jtitle>IEEE geoscience and remote sensing letters</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Theiler, James</au><au>Grosklos, Guen</au><aucorp>Los Alamos National Lab. (LANL), Los Alamos, NM (United States)</aucorp><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Problematic Projection to the In-Sample Subspace for a Kernelized Anomaly Detector</atitle><jtitle>IEEE geoscience and remote sensing letters</jtitle><stitle>LGRS</stitle><date>2016-04-01</date><risdate>2016</risdate><volume>13</volume><issue>4</issue><spage>485</spage><epage>489</epage><pages>485-489</pages><issn>1545-598X</issn><eissn>1558-0571</eissn><coden>IGRSBY</coden><abstract>We examine the properties and performance of kernelized anomaly detectors, with an emphasis on the Mahalanobis-distance-based kernel RX (KRX) algorithm. Although the detector generally performs well for high-bandwidth Gaussian kernels, it exhibits problematic (in some cases, catastrophic) performance for distances that are large compared to the bandwidth. By comparing KRX to two other anomaly detectors, we can trace the problem to a projection in feature space, which arises when a pseudoinverse is used on the covariance matrix in that feature space. We show that a regularized variant of KRX overcomes this difficulty and achieves superior performance over a wide range of bandwidths.</abstract><cop>Piscataway</cop><pub>IEEE</pub><doi>10.1109/LGRS.2016.2516985</doi><tpages>5</tpages><oa>free_for_read</oa></addata></record> |
| fulltext | fulltext |
| identifier | ISSN: 1545-598X |
| ispartof | IEEE geoscience and remote sensing letters, 2016-04, Vol.13 (4), p.485-489 |
| issn | 1545-598X 1558-0571 |
| language | eng |
| recordid | cdi_ieee_primary_7426793 |
| source | IEEE Electronic Library (IEL) Journals |
| subjects | Adaptive signal detection Algorithms Anomalies Anomaly detection Bandwidth Covariance matrices data models Detectors Gaussian Gaussian distribution GEOSCIENCES Hyperspectral imaging kernel density estimation kernel-RX Kernels mahalanobis distance MATHEMATICS AND COMPUTING multidimensional signal processing pattern recognition Projection Remote sensing singular value decomposition spectral analysis Unsupervised learning |
| title | Problematic Projection to the In-Sample Subspace for a Kernelized Anomaly Detector |
| url | http://sfxeu10.hosted.exlibrisgroup.com/loughborough?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-08T01%3A08%3A35IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest_ieee_&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Problematic%20Projection%20to%20the%20In-Sample%20Subspace%20for%20a%20Kernelized%20Anomaly%20Detector&rft.jtitle=IEEE%20geoscience%20and%20remote%20sensing%20letters&rft.au=Theiler,%20James&rft.aucorp=Los%20Alamos%20National%20Lab.%20(LANL),%20Los%20Alamos,%20NM%20(United%20States)&rft.date=2016-04-01&rft.volume=13&rft.issue=4&rft.spage=485&rft.epage=489&rft.pages=485-489&rft.issn=1545-598X&rft.eissn=1558-0571&rft.coden=IGRSBY&rft_id=info:doi/10.1109/LGRS.2016.2516985&rft_dat=%3Cproquest_ieee_%3E4046245741%3C/proquest_ieee_%3E%3Cgrp_id%3Ecdi_FETCH-LOGICAL-c353t-3e6617fee7336a105147bbde3cdbba645e6d29a6c9265e75eaccd6d212c46a6f3%3C/grp_id%3E%3Coa%3E%3C/oa%3E%3Curl%3E%3C/url%3E&rft_id=info:oai/&rft_pqid=1787028511&rft_id=info:pmid/&rft_ieee_id=7426793&rfr_iscdi=true |