Dynamic View-Adaptive Robust Representation Learning with Uncertainty-Aware Fusion

Dynamic View-Adaptive Robust Representation Learning with Uncertainty-Aware Fusion | 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 Research Article Dynamic View-Adaptive Robust Representation Learning with Uncertainty-Aware Fusion Emmanuel Ntaye, Xiang-Jun Shen, Ernest Domanaanmwi Ganaa, Conghu Zhou, and 2 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8025005/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Multi-view learning systems face significant reliability challenges in real-world applications due to sensor corruption, noise, and intermittent missing views.Current fusion strategies lack dynamic adaptation capabilities, compromising performance in safety-critical domains such as autonomous driving and medical diagnosis.We propose DAVE (Dynamic Adaptive View learning with Epistemic uncertainty), a robust multi-view framework that integrates uncertainty-aware stochastic encoders with a novel Uncertainty-Guided Adaptive Fusion (UGAF) module. Our approach dynamically weights view contributions based on real-time reliability estimates and incorporates robust training through stochastic view dropout and adversarial augmentation.Extensive evaluations across four diverse benchmarks demonstrate that DAVE achieves an average accuracy improvement of 8.6% under degraded conditions and reduces uncertainty miscalibration by 32% compared to state-of-the-art methods. The framework maintains robust performance even with 50% missing views and 40% sensor noise, establishing new standards for reliable multi-view learning.DAVE establishes a new paradigm for 1trustworthy multi-sensor systems by integrating principled uncertainty quantification with dynamic fusion. These advances enable reliable deployment in safety-critical applications where conventional multi-view approaches fail under real-world uncertainties. Theoretical Computer Science Multi-view learning uncertainty quantification adaptive fusion robust representation learning deep learning sensor fusion representation learning Full Text Additional Declarations The authors declare no competing interests. Cite Share Download PDF Status: Posted Version 1 posted 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-8025005","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":539582473,"identity":"4b110015-eaef-4391-93ff-e7ac50cc1f40","order_by":0,"name":"Emmanuel Ntaye","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA70lEQVRIiWNgGAWjYNCDBxUMDAYEVbEhcxLOkKwlsY0ILebyzc8kPlRsi+af3WP4IHHeYXlz9uYDDD8qtuHUYtnGZiY548zt3Bl3zhgbJG47bLiz51gCY8+Z2zi1GBxjMJPmbbud23Ajx0wCqIVxw40cA2bGNnxa2L9J//13O3f+jRzzH4lzDtsToYXHTJqx4XYuUKUZQ2LD4UQitOQUW/Ycu5278UZasUTCsfTkDWeOJRzE65fDxzfe+FFzO3fejeSNHz7UWNtuON588MGPCtxagIBFAonTDCYP4FMPBMwfkDh1BBSPglEwCkbBSAQAOnZjRrT4VKsAAAAASUVORK5CYII=","orcid":"https://orcid.org/0000-0001-6035-3834","institution":"Jiangsu University","correspondingAuthor":true,"prefix":"","firstName":"Emmanuel","middleName":"","lastName":"Ntaye","suffix":""},{"id":539582686,"identity":"b9e629ae-7ed8-4edd-84de-644501def481","order_by":1,"name":"Xiang-Jun Shen","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABFklEQVRIie3Rz0rDMBzA8V8JZJcfyzXi9g6/UvAPjO5VOgqeetDbhqCFXNWze4udck4pzMv2DoO+QGWXHkRNqjeNsptIvpAcUj6kSQBCoT/YVEVq16JMuVDGLeDQTmRHVHoIDdg6fhyd5UKuM7AI+a9E8OwYJ/NoWRbkCPDPL15yypAIC8nIbPb75yIdcUC66mAyXpnvyblC2smNHFJ9r2Wlc3QkQbhIfARqu0t853bZaqg0+yB2ffYTkbNXGa1M0bSVvu1J3MGbl1DNM2lQuuOD/bG6J4RgvORI2UsuUbpLPpFb/YSc8csEKU+WHiJEpZoXvHFP2bQLfT0VA6Xjbp6OH3xn+RoDTv1jHhLbHQhCoVDof/cO/WRZKMID7pMAAAAASUVORK5CYII=","orcid":"","institution":"Jiangsu University","correspondingAuthor":true,"prefix":"","firstName":"Xiang-Jun","middleName":"","lastName":"Shen","suffix":""},{"id":539582779,"identity":"bcbb8897-83e2-4396-95fd-b2bb5a704c06","order_by":2,"name":"Ernest Domanaanmwi Ganaa","email":"","orcid":"","institution":"Hilla Limann Technical University","correspondingAuthor":false,"prefix":"","firstName":"Ernest","middleName":"Domanaanmwi","lastName":"Ganaa","suffix":""},{"id":539583218,"identity":"688f954a-a9cb-43e2-aed4-8875353de03c","order_by":3,"name":"Conghu Zhou","email":"","orcid":"","institution":"Jiangsu University","correspondingAuthor":false,"prefix":"","firstName":"Conghu","middleName":"","lastName":"Zhou","suffix":""},{"id":539583219,"identity":"ca73020c-c52b-4838-b81b-2fc7e51b17ad","order_by":4,"name":"Zhifeng Liu","email":"","orcid":"","institution":"Jiangsu University","correspondingAuthor":false,"prefix":"","firstName":"Zhifeng","middleName":"","lastName":"Liu","suffix":""},{"id":539583220,"identity":"a5404906-5de8-4b04-ad6d-c583c91084d6","order_by":5,"name":"Heping Song","email":"","orcid":"","institution":"Jiangsu University","correspondingAuthor":false,"prefix":"","firstName":"Heping","middleName":"","lastName":"Song","suffix":""}],"badges":[],"createdAt":"2025-11-04 06:11:10","currentVersionCode":1,"declarations":{"humanSubjects":false,"vertebrateSubjects":false,"conflictsOfInterestStatement":false,"humanSubjectEthicalGuidelines":false,"humanSubjectConsent":false,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":false},"doi":"10.21203/rs.3.rs-8025005/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8025005/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":95227064,"identity":"fe3a98ad-6483-4a90-a94f-92db85c7b917","added_by":"auto","created_at":"2025-11-05 16:32:02","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":4760708,"visible":true,"origin":"","legend":"","description":"","filename":"DVRLUAWF.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8025005/v1_covered_f57785e2-309f-4186-a7a2-4afe8f7e3d91.pdf"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"\u003cp\u003eDynamic View-Adaptive Robust Representation Learning with Uncertainty-Aware Fusion\u003c/p\u003e","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"Jiangsu University","isAcceptedByJournal":false,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Multi-view learning, uncertainty quantification, adaptive fusion, robust representation learning, deep learning, sensor fusion, representation learning","lastPublishedDoi":"10.21203/rs.3.rs-8025005/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8025005/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eMulti-view learning systems face significant reliability challenges in real-world applications due to sensor \u0026nbsp;corruption, noise, and intermittent missing views.Current fusion strategies lack dynamic adaptation capabilities, compromising performance in safety-critical domains such as autonomous driving and medical diagnosis.We propose DAVE (Dynamic Adaptive View learning with Epistemic uncertainty), a robust multi-view framework that integrates uncertainty-aware stochastic encoders with a novel Uncertainty-Guided Adaptive Fusion (UGAF) module. Our approach dynamically weights view contributions based on real-time reliability estimates and incorporates robust training through stochastic view dropout and adversarial augmentation.Extensive evaluations across four diverse benchmarks demonstrate that DAVE achieves an average accuracy improvement of 8.6% under degraded conditions and reduces uncertainty miscalibration by 32% compared to state-of-the-art methods. The framework maintains robust performance even with 50% missing views and 40% sensor noise, establishing new standards for reliable multi-view learning.DAVE establishes a new paradigm for 1trustworthy multi-sensor systems by integrating principled uncertainty quantification with dynamic fusion. These advances enable reliable deployment in safety-critical applications where conventional multi-view approaches fail under real-world uncertainties.\u003c/p\u003e","manuscriptTitle":"Dynamic View-Adaptive Robust Representation Learning with Uncertainty-Aware Fusion","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-11-05 05:04:25","doi":"10.21203/rs.3.rs-8025005/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"027fca34-00be-4717-af40-1b502c03fd80","owner":[],"postedDate":"November 5th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":57387629,"name":"Theoretical Computer Science"}],"tags":[],"updatedAt":"2025-11-05T05:04:25+00:00","versionOfRecord":[],"versionCreatedAt":"2025-11-05 05:04:25","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8025005","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8025005","identity":"rs-8025005","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}