Geometric-Aware Representation Learning for Semi-Supervised Domain Generalization

Geometric-Aware Representation Learning for Semi-Supervised Domain Generalization | 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 Geometric-Aware Representation Learning for Semi-Supervised Domain Generalization Ali Atghaei, Mohammad Rahmati This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9662991/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 Semi-supervised domain generalization (SSDG) aims to learn models that generalize effectively to unseen domains using limited labeled source data. Existing methods often overlook the geometric structure of feature representations across domains and augmentations, which limits robustness and cross-domain consistency. In this work, we propose a geometry-aware framework that integrates graph-based label propagation and Nyström-based feature alignment to refine pseudo-labels and preserve structural consistency in the learned feature space. Extensive experiments conducted on the PACS, OfficeHome, VLCS, and Digits-DG benchmarks demonstrate that the proposed method consistently outperforms existing state-of-the-art approaches, achieving an average improvement of 2.56% on PACS. These results highlight the effectiveness of incorporating geometric constraints into semi-supervised domain generalization under limited-label settings. To promote transparency and reproducibility, the official implementation, experimental configurations, and evaluation scripts are publicly available at: https://github.com/ali-atghaei/geo_ssdg. Artificial Intelligence and Machine Learning Domain Generalization Domain Adaptation Semi-Supervised Learning Label Propagation 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-9662991","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":637338186,"identity":"7aed04ec-9a41-4b1d-9e72-ee8b01c0ad8c","order_by":0,"name":"Ali Atghaei","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAyklEQVRIiWNgGAWjYHACNiCWkIPxeIjRANZizMDATJoWhsQGqBbCgH9++7UHPyos0vtn5B9g+FHDIGPeQECLxDGecsOeMxK5M24kMzD2HGPgkTlAyJpjPGkSvG0SuQ1ALQy8DQw8EoR0yAO1SP79J5EuD7LlLzFaDI6xH5PmbZBIMABqYSbKFsNjOWzSMsckDDeeeWxwGMggrEXu8PFnkm9q6uTljic+fPimxsaeoBZg3BnAmQeAIUhYAwMD+wNiVI2CUTAKRsFIBgCMLTY5WGzmugAAAABJRU5ErkJggg==","orcid":"https://orcid.org/0000-0001-8630-9153","institution":"Amirkabir university of technology","correspondingAuthor":true,"prefix":"","firstName":"Ali","middleName":"","lastName":"Atghaei","suffix":""},{"id":637338447,"identity":"b4eff3db-389a-4ed5-a208-18a3e8594f93","order_by":1,"name":"Mohammad Rahmati","email":"","orcid":"https://orcid.org/0000-0002-0591-6910","institution":"Amirkabir university of technology","correspondingAuthor":false,"prefix":"","firstName":"Mohammad","middleName":"","lastName":"Rahmati","suffix":""}],"badges":[],"createdAt":"2026-05-09 11:08:05","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-9662991/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9662991/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":109204999,"identity":"4eca0a30-afe7-457a-a128-068d875ee275","added_by":"auto","created_at":"2026-05-13 15:03:09","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2876070,"visible":true,"origin":"","legend":"","description":"","filename":"SpringerNatureLaTeXTemplate2.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9662991/v1_covered_352f8ae9-84db-463e-9963-ea850f545725.pdf"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"\u003cp\u003eGeometric-Aware Representation Learning for Semi-Supervised Domain Generalization\u003c/p\u003e","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"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":"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":"Domain Generalization, Domain Adaptation, Semi-Supervised Learning, Label Propagation","lastPublishedDoi":"10.21203/rs.3.rs-9662991/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9662991/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eSemi-supervised domain generalization (SSDG) aims to learn models that generalize effectively to unseen domains using limited labeled source data. Existing methods often overlook the geometric structure of feature representations across domains and augmentations, which limits robustness and cross-domain consistency. In this work, we propose a geometry-aware framework that integrates graph-based label propagation and Nyström-based feature alignment to refine pseudo-labels and preserve structural consistency in the learned feature space. Extensive experiments conducted on the PACS, OfficeHome, VLCS, and Digits-DG benchmarks demonstrate that the proposed method consistently outperforms existing state-of-the-art approaches, achieving an average improvement of \u003cstrong\u003e2.56%\u003c/strong\u003e on PACS. These results highlight the effectiveness of incorporating geometric constraints into semi-supervised domain generalization under limited-label settings. To promote transparency and reproducibility, the official implementation, experimental configurations, and evaluation scripts are publicly available at: https://github.com/ali-atghaei/geo_ssdg.\u003c/p\u003e","manuscriptTitle":"Geometric-Aware Representation Learning for Semi-Supervised Domain Generalization","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-05-12 19:01:03","doi":"10.21203/rs.3.rs-9662991/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":"22a8e565-f4d1-404e-8a8e-f4260166e9eb","owner":[],"postedDate":"May 12th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":67837482,"name":"Artificial Intelligence and Machine Learning"}],"tags":[],"updatedAt":"2026-05-12T19:01:03+00:00","versionOfRecord":[],"versionCreatedAt":"2026-05-12 19:01:03","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9662991","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9662991","identity":"rs-9662991","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}