MemCLR-GANomaly: Unsupervised Anomaly Detection with Dynamic Memory Bank and Contrastive Learning

MemCLR-GANomaly: Unsupervised Anomaly Detection with Dynamic Memory Bank and Contrastive Learning | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article MemCLR-GANomaly: Unsupervised Anomaly Detection with Dynamic Memory Bank and Contrastive Learning Xiaona Song, Runqing Zhang, Lijun Wang, Kaixuan Lv, Ying Zhu This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9293833/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 12 You are reading this latest preprint version Abstract Anomaly detection in complex industrial scenarios requires models that characterize normal patterns and maintain high sensitivity to subtle anomalies. Generative Adversarial Networks (GANs) have attracted increasing attention for their strong representation and generation capabilities. Among them, GANomaly adopts an encoder-decoder-encoder architecture and computes anomaly scores via latent vector discrepancies, providing a solid baseline. However, its scoring mechanism remains limited in capturing pixel-, feature-, and distribution-level anomalies. To address this, we propose MemCLR-GANomaly, which integrates multiple collaborative modules. We introduce SimCLR-based contrastive pre-training for better feature discriminability, a multi-scale region error module for local anomaly perception, and a dynamic memory bank to model normal distributions. An adaptive fusion network integrates these indicators. On MVTec AD, our method achieves 0.919 mean AUC, best in 13/15 categories, improving over GANomaly and Skip-GANomaly by 25.9% and 9.2%, respectively. On a custom Grinding Wheel dataset, it achieves 0.982 AUC, improving over Skip-GANomaly by 9.4%. Physical sciences/Engineering Physical sciences/Mathematics and computing Anomaly Detection SimCLR Dynamic Memory Bank Adaptive Fusion Network Full Text Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Editorial decision: Revision requested 04 May, 2026 Reviews received at journal 02 May, 2026 Reviews received at journal 01 May, 2026 Reviewers agreed at journal 29 Apr, 2026 Reviewers agreed at journal 22 Apr, 2026 Reviews received at journal 20 Apr, 2026 Reviewers agreed at journal 19 Apr, 2026 Reviewers invited by journal 15 Apr, 2026 Editor assigned by journal 15 Apr, 2026 Editor invited by journal 15 Apr, 2026 Submission checks completed at journal 11 Apr, 2026 First submitted to journal 11 Apr, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-9293833","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":626042719,"identity":"613e115a-3839-47a1-a05e-2796e2e35440","order_by":0,"name":"Xiaona Song","email":"","orcid":"","institution":"North China University of Water Resources and Electric Power","correspondingAuthor":false,"prefix":"","firstName":"Xiaona","middleName":"","lastName":"Song","suffix":""},{"id":626042723,"identity":"4089578a-3ccd-4c6a-9be3-d9f361dec07c","order_by":1,"name":"Runqing Zhang","email":"data:image/png;base64,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","orcid":"","institution":"North China University of Water Resources and Electric Power","correspondingAuthor":true,"prefix":"","firstName":"Runqing","middleName":"","lastName":"Zhang","suffix":""},{"id":626042726,"identity":"27435638-5460-41d7-bc21-2960e392f390","order_by":2,"name":"Lijun Wang","email":"","orcid":"","institution":"North China University of Water Resources and Electric Power","correspondingAuthor":false,"prefix":"","firstName":"Lijun","middleName":"","lastName":"Wang","suffix":""},{"id":626042730,"identity":"d5f85b90-e986-48a1-adce-73d9688c112c","order_by":3,"name":"Kaixuan Lv","email":"","orcid":"","institution":"North China University of Water Resources and Electric Power","correspondingAuthor":false,"prefix":"","firstName":"Kaixuan","middleName":"","lastName":"Lv","suffix":""},{"id":626042733,"identity":"1e944355-3037-4f4f-8a90-b3959490ee53","order_by":4,"name":"Ying Zhu","email":"","orcid":"","institution":"China Railway Engineering Equipment Group Co., Ltd.","correspondingAuthor":false,"prefix":"","firstName":"Ying","middleName":"","lastName":"Zhu","suffix":""}],"badges":[],"createdAt":"2026-04-01 15:24:50","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9293833/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9293833/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":107490261,"identity":"50b5af9a-b16a-4627-b2a4-beb9d97cdb0a","added_by":"auto","created_at":"2026-04-22 02:51:28","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":924211,"visible":true,"origin":"","legend":"","description":"","filename":"TemplateforsubmissionstoScientificReports.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9293833/v1_covered_fda8960f-c5af-4fca-8e6e-faf71af415c7.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"MemCLR-GANomaly: Unsupervised Anomaly Detection with Dynamic Memory Bank and Contrastive Learning","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Anomaly Detection, SimCLR, Dynamic Memory Bank, Adaptive Fusion Network","lastPublishedDoi":"10.21203/rs.3.rs-9293833/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9293833/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Anomaly detection in complex industrial scenarios requires models that characterize normal patterns and maintain high sensitivity to subtle anomalies. Generative Adversarial Networks (GANs) have attracted increasing attention for their strong representation and generation capabilities. Among them, GANomaly adopts an encoder-decoder-encoder architecture and computes anomaly scores via latent vector discrepancies, providing a solid baseline. However, its scoring mechanism remains limited in capturing pixel-, feature-, and distribution-level anomalies. To address this, we propose MemCLR-GANomaly, which integrates multiple collaborative modules. We introduce SimCLR-based contrastive pre-training for better feature discriminability, a multi-scale region error module for local anomaly perception, and a dynamic memory bank to model normal distributions. An adaptive fusion network integrates these indicators. On MVTec AD, our method achieves 0.919 mean AUC, best in 13/15 categories, improving over GANomaly and Skip-GANomaly by 25.9% and 9.2%, respectively. On a custom Grinding Wheel dataset, it achieves 0.982 AUC, improving over Skip-GANomaly by 9.4%.","manuscriptTitle":"MemCLR-GANomaly: Unsupervised Anomaly Detection with Dynamic Memory Bank and Contrastive Learning","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-04-21 14:47:44","doi":"10.21203/rs.3.rs-9293833/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2026-05-04T12:06:28+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-05-02T08:20:26+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-05-01T08:47:16+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"189763757527710053624796902069086258311","date":"2026-04-29T11:09:01+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"241134625904492309322979997284650024842","date":"2026-04-22T09:51:22+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-04-20T06:22:41+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"264631290348744657139609465069093797697","date":"2026-04-20T03:59:05+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-04-15T20:06:33+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-04-15T20:04:38+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2026-04-15T17:01:00+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-04-11T06:57:15+00:00","index":"","fulltext":""},{"type":"submitted","content":"Scientific Reports","date":"2026-04-11T06:54:55+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"9a5d7de6-083b-4242-81bd-51e04adbe54f","owner":[],"postedDate":"April 21st, 2026","published":true,"recentEditorialEvents":[{"type":"decision","content":"Revision requested","date":"2026-05-04T12:06:28+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-05-02T08:20:26+00:00","index":55,"fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-05-01T08:47:16+00:00","index":53,"fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":66748408,"name":"Physical sciences/Engineering"},{"id":66748409,"name":"Physical sciences/Mathematics and computing"}],"tags":[],"updatedAt":"2026-05-15T06:54:36+00:00","versionOfRecord":[],"versionCreatedAt":"2026-04-21 14:47:44","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9293833","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9293833","identity":"rs-9293833","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}