Quickest Change Detection for Unnormalized Statistical Models
Classical quickest change detection algorithms require modeling pre-change and post-change distributions. Such an approach may not be feasible for various machine learning models because of the complexity of computing the explicit distributions. Additionally, these methods may suffer from a lack of...
Gespeichert in:
| Veröffentlicht in: | IEEE transactions on information theory 2024-02, Vol.70 (2), p.1-1 |
|---|---|
| Hauptverfasser: | , , , , |
| Format: | Artikel |
| Sprache: | eng |
| Schlagworte: | |
| Online-Zugang: | Volltext bestellen |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| container_end_page | 1 |
|---|---|
| container_issue | 2 |
| container_start_page | 1 |
| container_title | IEEE transactions on information theory |
| container_volume | 70 |
| creator | Wu, Suya Diao, Enmao Banerjee, Taposh Ding, Jie Tarokh, Vahid |
| description | Classical quickest change detection algorithms require modeling pre-change and post-change distributions. Such an approach may not be feasible for various machine learning models because of the complexity of computing the explicit distributions. Additionally, these methods may suffer from a lack of robustness to model mismatch and noise. This paper develops a new variant of the classical Cumulative Sum (CUSUM) algorithm for the quickest change detection. This variant is based on Fisher divergence and the Hyvärinen score and is called the Hyvärinen score-based CUSUM (SCUSUM) algorithm. The SCUSUM algorithm allows the applications of change detection for unnormalized statistical models, i.e., models for which the probability density function contains an unknown normalization constant. The asymptotic optimality of the proposed algorithm is investigated by deriving expressions for average detection delay and the mean running time to a false alarm. Numerical results are provided to demonstrate the performance of the proposed algorithm. |
| doi_str_mv | 10.1109/TIT.2023.3328274 |
| format | Article |
| fullrecord | <record><control><sourceid>proquest_RIE</sourceid><recordid>TN_cdi_ieee_primary_10298279</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><ieee_id>10298279</ieee_id><sourcerecordid>2917937052</sourcerecordid><originalsourceid>FETCH-LOGICAL-c245t-537b7350f8357ef65ff26d78e483ddf7a06833951d7243926f9f9d8e14648d913</originalsourceid><addsrcrecordid>eNpNkD1PAzEMhiMEEqWwMzCcxHwln5dkYEDlq1IRQrRzFC4OXLleIEkH-PWkagcmy_Lz2taD0DnBE0KwvlrMFhOKKZswRhWV_ACNiBCy1o3gh2iEMVG15lwdo5OUVqXlgtARun7ZdO0npFxNP-zwDtUtZGhzF4bKh1gthyHEte27X3DVa7a5S7lrbV89BQd9OkVH3vYJzvZ1jJb3d4vpYz1_fphNb-Z1S7nItWDyTTKBvWJCgm-E97RxUgFXzDkvLW4UY1oQJylnmjZee-0UEN5w5TRhY3S52_sVw_emfGtWYROHctJQTaRmEgtaKLyj2hhSiuDNV-zWNv4Ygs1WkimSzFaS2UsqkYtdpAOAfzjVZazZH2tQYWY</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2917937052</pqid></control><display><type>article</type><title>Quickest Change Detection for Unnormalized Statistical Models</title><source>IEEE Electronic Library (IEL)</source><creator>Wu, Suya ; Diao, Enmao ; Banerjee, Taposh ; Ding, Jie ; Tarokh, Vahid</creator><creatorcontrib>Wu, Suya ; Diao, Enmao ; Banerjee, Taposh ; Ding, Jie ; Tarokh, Vahid</creatorcontrib><description>Classical quickest change detection algorithms require modeling pre-change and post-change distributions. Such an approach may not be feasible for various machine learning models because of the complexity of computing the explicit distributions. Additionally, these methods may suffer from a lack of robustness to model mismatch and noise. This paper develops a new variant of the classical Cumulative Sum (CUSUM) algorithm for the quickest change detection. This variant is based on Fisher divergence and the Hyvärinen score and is called the Hyvärinen score-based CUSUM (SCUSUM) algorithm. The SCUSUM algorithm allows the applications of change detection for unnormalized statistical models, i.e., models for which the probability density function contains an unknown normalization constant. The asymptotic optimality of the proposed algorithm is investigated by deriving expressions for average detection delay and the mean running time to a false alarm. Numerical results are provided to demonstrate the performance of the proposed algorithm.</description><identifier>ISSN: 0018-9448</identifier><identifier>EISSN: 1557-9654</identifier><identifier>DOI: 10.1109/TIT.2023.3328274</identifier><identifier>CODEN: IETTAW</identifier><language>eng</language><publisher>New York: IEEE</publisher><subject>Algorithms ; Change detection ; Change detection algorithms ; Computational modeling ; CUSUM ; Delays ; Detection algorithms ; Divergence ; False alarms ; Fisher divergence ; Machine learning ; Numerical models ; Probability density functions ; Quickest change detection ; Random variables ; Robustness (mathematics) ; Score matching ; Statistical analysis ; Statistical models ; Unnormalized models</subject><ispartof>IEEE transactions on information theory, 2024-02, Vol.70 (2), p.1-1</ispartof><rights>Copyright The Institute of Electrical and Electronics Engineers, Inc. (IEEE) 2024</rights><lds50>peer_reviewed</lds50><woscitedreferencessubscribed>false</woscitedreferencessubscribed><cites>FETCH-LOGICAL-c245t-537b7350f8357ef65ff26d78e483ddf7a06833951d7243926f9f9d8e14648d913</cites><orcidid>0000-0003-4163-6913 ; 0000-0002-3584-6140 ; 0000-0002-5383-0900 ; 0000-0003-2994-6302</orcidid></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktohtml>$$Uhttps://ieeexplore.ieee.org/document/10298279$$EHTML$$P50$$Gieee$$H</linktohtml><link.rule.ids>314,776,780,792,27903,27904,54736</link.rule.ids><linktorsrc>$$Uhttps://ieeexplore.ieee.org/document/10298279$$EView_record_in_IEEE$$FView_record_in_$$GIEEE</linktorsrc></links><search><creatorcontrib>Wu, Suya</creatorcontrib><creatorcontrib>Diao, Enmao</creatorcontrib><creatorcontrib>Banerjee, Taposh</creatorcontrib><creatorcontrib>Ding, Jie</creatorcontrib><creatorcontrib>Tarokh, Vahid</creatorcontrib><title>Quickest Change Detection for Unnormalized Statistical Models</title><title>IEEE transactions on information theory</title><addtitle>TIT</addtitle><description>Classical quickest change detection algorithms require modeling pre-change and post-change distributions. Such an approach may not be feasible for various machine learning models because of the complexity of computing the explicit distributions. Additionally, these methods may suffer from a lack of robustness to model mismatch and noise. This paper develops a new variant of the classical Cumulative Sum (CUSUM) algorithm for the quickest change detection. This variant is based on Fisher divergence and the Hyvärinen score and is called the Hyvärinen score-based CUSUM (SCUSUM) algorithm. The SCUSUM algorithm allows the applications of change detection for unnormalized statistical models, i.e., models for which the probability density function contains an unknown normalization constant. The asymptotic optimality of the proposed algorithm is investigated by deriving expressions for average detection delay and the mean running time to a false alarm. Numerical results are provided to demonstrate the performance of the proposed algorithm.</description><subject>Algorithms</subject><subject>Change detection</subject><subject>Change detection algorithms</subject><subject>Computational modeling</subject><subject>CUSUM</subject><subject>Delays</subject><subject>Detection algorithms</subject><subject>Divergence</subject><subject>False alarms</subject><subject>Fisher divergence</subject><subject>Machine learning</subject><subject>Numerical models</subject><subject>Probability density functions</subject><subject>Quickest change detection</subject><subject>Random variables</subject><subject>Robustness (mathematics)</subject><subject>Score matching</subject><subject>Statistical analysis</subject><subject>Statistical models</subject><subject>Unnormalized models</subject><issn>0018-9448</issn><issn>1557-9654</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2024</creationdate><recordtype>article</recordtype><sourceid>RIE</sourceid><recordid>eNpNkD1PAzEMhiMEEqWwMzCcxHwln5dkYEDlq1IRQrRzFC4OXLleIEkH-PWkagcmy_Lz2taD0DnBE0KwvlrMFhOKKZswRhWV_ACNiBCy1o3gh2iEMVG15lwdo5OUVqXlgtARun7ZdO0npFxNP-zwDtUtZGhzF4bKh1gthyHEte27X3DVa7a5S7lrbV89BQd9OkVH3vYJzvZ1jJb3d4vpYz1_fphNb-Z1S7nItWDyTTKBvWJCgm-E97RxUgFXzDkvLW4UY1oQJylnmjZee-0UEN5w5TRhY3S52_sVw_emfGtWYROHctJQTaRmEgtaKLyj2hhSiuDNV-zWNv4Ygs1WkimSzFaS2UsqkYtdpAOAfzjVZazZH2tQYWY</recordid><startdate>20240201</startdate><enddate>20240201</enddate><creator>Wu, Suya</creator><creator>Diao, Enmao</creator><creator>Banerjee, Taposh</creator><creator>Ding, Jie</creator><creator>Tarokh, Vahid</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>AAYXX</scope><scope>CITATION</scope><scope>7SC</scope><scope>7SP</scope><scope>8FD</scope><scope>JQ2</scope><scope>L7M</scope><scope>L~C</scope><scope>L~D</scope><orcidid>https://orcid.org/0000-0003-4163-6913</orcidid><orcidid>https://orcid.org/0000-0002-3584-6140</orcidid><orcidid>https://orcid.org/0000-0002-5383-0900</orcidid><orcidid>https://orcid.org/0000-0003-2994-6302</orcidid></search><sort><creationdate>20240201</creationdate><title>Quickest Change Detection for Unnormalized Statistical Models</title><author>Wu, Suya ; Diao, Enmao ; Banerjee, Taposh ; Ding, Jie ; Tarokh, Vahid</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c245t-537b7350f8357ef65ff26d78e483ddf7a06833951d7243926f9f9d8e14648d913</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2024</creationdate><topic>Algorithms</topic><topic>Change detection</topic><topic>Change detection algorithms</topic><topic>Computational modeling</topic><topic>CUSUM</topic><topic>Delays</topic><topic>Detection algorithms</topic><topic>Divergence</topic><topic>False alarms</topic><topic>Fisher divergence</topic><topic>Machine learning</topic><topic>Numerical models</topic><topic>Probability density functions</topic><topic>Quickest change detection</topic><topic>Random variables</topic><topic>Robustness (mathematics)</topic><topic>Score matching</topic><topic>Statistical analysis</topic><topic>Statistical models</topic><topic>Unnormalized models</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Wu, Suya</creatorcontrib><creatorcontrib>Diao, Enmao</creatorcontrib><creatorcontrib>Banerjee, Taposh</creatorcontrib><creatorcontrib>Ding, Jie</creatorcontrib><creatorcontrib>Tarokh, Vahid</creatorcontrib><collection>IEEE All-Society Periodicals Package (ASPP) 2005-present</collection><collection>IEEE All-Society Periodicals Package (ASPP) 1998-Present</collection><collection>IEEE Electronic Library (IEL)</collection><collection>CrossRef</collection><collection>Computer and Information Systems Abstracts</collection><collection>Electronics & Communications 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>IEEE transactions on information theory</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Wu, Suya</au><au>Diao, Enmao</au><au>Banerjee, Taposh</au><au>Ding, Jie</au><au>Tarokh, Vahid</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Quickest Change Detection for Unnormalized Statistical Models</atitle><jtitle>IEEE transactions on information theory</jtitle><stitle>TIT</stitle><date>2024-02-01</date><risdate>2024</risdate><volume>70</volume><issue>2</issue><spage>1</spage><epage>1</epage><pages>1-1</pages><issn>0018-9448</issn><eissn>1557-9654</eissn><coden>IETTAW</coden><abstract>Classical quickest change detection algorithms require modeling pre-change and post-change distributions. Such an approach may not be feasible for various machine learning models because of the complexity of computing the explicit distributions. Additionally, these methods may suffer from a lack of robustness to model mismatch and noise. This paper develops a new variant of the classical Cumulative Sum (CUSUM) algorithm for the quickest change detection. This variant is based on Fisher divergence and the Hyvärinen score and is called the Hyvärinen score-based CUSUM (SCUSUM) algorithm. The SCUSUM algorithm allows the applications of change detection for unnormalized statistical models, i.e., models for which the probability density function contains an unknown normalization constant. The asymptotic optimality of the proposed algorithm is investigated by deriving expressions for average detection delay and the mean running time to a false alarm. Numerical results are provided to demonstrate the performance of the proposed algorithm.</abstract><cop>New York</cop><pub>IEEE</pub><doi>10.1109/TIT.2023.3328274</doi><tpages>1</tpages><orcidid>https://orcid.org/0000-0003-4163-6913</orcidid><orcidid>https://orcid.org/0000-0002-3584-6140</orcidid><orcidid>https://orcid.org/0000-0002-5383-0900</orcidid><orcidid>https://orcid.org/0000-0003-2994-6302</orcidid></addata></record> |
| fulltext | fulltext_linktorsrc |
| identifier | ISSN: 0018-9448 |
| ispartof | IEEE transactions on information theory, 2024-02, Vol.70 (2), p.1-1 |
| issn | 0018-9448 1557-9654 |
| language | eng |
| recordid | cdi_ieee_primary_10298279 |
| source | IEEE Electronic Library (IEL) |
| subjects | Algorithms Change detection Change detection algorithms Computational modeling CUSUM Delays Detection algorithms Divergence False alarms Fisher divergence Machine learning Numerical models Probability density functions Quickest change detection Random variables Robustness (mathematics) Score matching Statistical analysis Statistical models Unnormalized models |
| title | Quickest Change Detection for Unnormalized Statistical Models |
| url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-27T07%3A43%3A18IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest_RIE&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Quickest%20Change%20Detection%20for%20Unnormalized%20Statistical%20Models&rft.jtitle=IEEE%20transactions%20on%20information%20theory&rft.au=Wu,%20Suya&rft.date=2024-02-01&rft.volume=70&rft.issue=2&rft.spage=1&rft.epage=1&rft.pages=1-1&rft.issn=0018-9448&rft.eissn=1557-9654&rft.coden=IETTAW&rft_id=info:doi/10.1109/TIT.2023.3328274&rft_dat=%3Cproquest_RIE%3E2917937052%3C/proquest_RIE%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=2917937052&rft_id=info:pmid/&rft_ieee_id=10298279&rfr_iscdi=true |