ResNet34-Based Galaxy Morphology Classification with Machine Unlearning

ResNet34-Based Galaxy Morphology Classification with Machine Unlearning | 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 ResNet34-Based Galaxy Morphology Classification with Machine Unlearning Nayna Potdukhe, Priyani Sabde, Pooja Aaglave, Meet Upadhye, Smit Barmate, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9269285/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 Galaxy morphology classification is fundamental to observational astronomy. The structure of a galaxy, whether it is smooth and elliptical, features spiral arms and rotation, or presents an optical artifact—tells us a lot about how it formed, if it has merged with others, and where it is headed evolutionarily. As surveys have grown from thousands to hundreds of thousands of galaxies, manually classifying each one has become impossible. Automated deep-learning pipelines have therefore become the standard approach for scaling galaxy morphology classification. However, a challenge arises. Large citizen-science datasets like Galaxy Zoo 2 suffer from label noise because volunteers have different expertise levels and sometimes disagree on ambiguous images. This noise is especially damaging for rare classes; a small percentage of wrong labels can significantly compromise what the model learns. Once a model is trained on noisy data, standard fine-tuning does not provide an effective method to fix labeled errors without retraining entirely. This paper addresses both problems simultaneously. We trained a ResNet34-based CNN on 61,578 Galaxy Zoo 2 images to classify galaxies into three categories: Smooth, Featured/Disk, and Artifact, then applied three machine unlearning methods to reduce the influence of approximately 10–12% intentionally mislabeled samples. We compared Gradient Ascent Unlearning, Fisher Forgetting, and Full Retraining. Our classifier achieved 85.58% validation accuracy, 79.34% balanced accuracy, and a macro-F1 score of 75.15%. Among unlearning methods, Gradient Ascent was fastest (11.01% forget-set accuracy in 103.6 seconds), while Full Retraining gave the best retention (99.64% at 2649.4 seconds). Our experiments also reveal an important implementation constraint: Fisher Forgetting can collapse when the numerical stabilizer is not large enough. Parameters with low importance receive substantial noise perturbations instead of selective forgetting. This is a non-trivial issue with significant practical implications. Machine Unlearning Galaxy Morphology Classification ResNet34 Galaxy Zoo 2 Noisy Labels Deep Learning Astronomical Image Analysis Full Text Additional Declarations No competing interests reported. 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-9269285","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":618371164,"identity":"253221f0-ec99-4617-adf7-671455378f1d","order_by":0,"name":"Nayna Potdukhe","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA8klEQVRIiWNgGAWjYFCDA2xsDAwVQAYzcwNRGiQgWs6AtDCSooWxDcQmoMWc/YzZg587bOr4bqSlPfg4rzaavx2o5UfFNpxaLHtyzA17z6RJSN5IO244c9vx3BmHGRsYe87cxqnF4ECOmQRv22EJgxvpbdK8247lNgC1MDO24dFy/o2Z5F+Ylr9zjuXOJ6jlRo6ZNMSWtGPSjA01uRsIa3lWJi17Jk1y5plnaZI9xw7kbgRqOYjXL+eTt0m+3WHDz3c8zUziR01d7rzzhw8++FGBWwsDA4cBckQcBpMH8KgHAvYHyFrq8CseBaNgFIyCEQkA/x9gptw6LA4AAAAASUVORK5CYII=","orcid":"","institution":"Government Polytechnic","correspondingAuthor":true,"prefix":"","firstName":"Nayna","middleName":"","lastName":"Potdukhe","suffix":""},{"id":618371166,"identity":"3acde4e9-3aa4-4e89-80d3-9766306eaae4","order_by":1,"name":"Priyani Sabde","email":"","orcid":"","institution":"Government Polytechnic","correspondingAuthor":false,"prefix":"","firstName":"Priyani","middleName":"","lastName":"Sabde","suffix":""},{"id":618371169,"identity":"e5f1da97-903d-4b90-9799-7a8d29286f18","order_by":2,"name":"Pooja Aaglave","email":"","orcid":"","institution":"Government Polytechnic","correspondingAuthor":false,"prefix":"","firstName":"Pooja","middleName":"","lastName":"Aaglave","suffix":""},{"id":618371170,"identity":"f4c48440-ccf2-41f9-9262-e9c0cd89c0d9","order_by":3,"name":"Meet Upadhye","email":"","orcid":"","institution":"Government Polytechnic","correspondingAuthor":false,"prefix":"","firstName":"Meet","middleName":"","lastName":"Upadhye","suffix":""},{"id":618371171,"identity":"904605f4-8dbc-45b6-af90-01982b7f2026","order_by":4,"name":"Smit Barmate","email":"","orcid":"","institution":"Government Polytechnic","correspondingAuthor":false,"prefix":"","firstName":"Smit","middleName":"","lastName":"Barmate","suffix":""},{"id":618371172,"identity":"bdbd6112-822e-4d10-b5ad-380612d9faf6","order_by":5,"name":"Siddhesh Ratnaparkhi","email":"","orcid":"","institution":"Government Polytechnic","correspondingAuthor":false,"prefix":"","firstName":"Siddhesh","middleName":"","lastName":"Ratnaparkhi","suffix":""}],"badges":[],"createdAt":"2026-03-30 16:00:11","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9269285/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9269285/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":106959036,"identity":"1915105a-dd29-40e5-8c2a-3870f4d5b720","added_by":"auto","created_at":"2026-04-15 08:44:11","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":666984,"visible":true,"origin":"","legend":"","description":"","filename":"resnet34basedgalaxymorphologyclassificationwithmachineunlearning1.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9269285/v1_covered_ec54f4eb-7161-4965-a07b-f2e87085ab1e.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"ResNet34-Based Galaxy Morphology Classification with Machine Unlearning","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":"Machine Unlearning, Galaxy Morphology Classification, ResNet34, Galaxy Zoo 2, Noisy Labels, Deep Learning, Astronomical Image Analysis","lastPublishedDoi":"10.21203/rs.3.rs-9269285/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9269285/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eGalaxy morphology classification is fundamental to observational astronomy. The structure of a galaxy, whether it is smooth and elliptical, features spiral arms and rotation, or presents an optical artifact\u0026mdash;tells us a lot about how it formed, if it has merged with others, and where it is headed evolutionarily. As surveys have grown from thousands to hundreds of thousands of galaxies, manually classifying each one has become impossible. Automated deep-learning pipelines have therefore become the standard approach for scaling galaxy morphology classification. However, a challenge arises. Large citizen-science datasets like Galaxy Zoo 2 suffer from label noise because volunteers have different expertise levels and sometimes disagree on ambiguous images. This noise is especially damaging for rare classes; a small percentage of wrong labels can significantly compromise what the model learns. Once a model is trained on noisy data, standard fine-tuning does not provide an effective method to fix labeled errors without retraining entirely.\u003c/p\u003e \u003cp\u003eThis paper addresses both problems simultaneously. We trained a ResNet34-based CNN on 61,578 Galaxy Zoo 2 images to classify galaxies into three categories: Smooth, Featured/Disk, and Artifact, then applied three machine unlearning methods to reduce the influence of approximately 10\u0026ndash;12% intentionally mislabeled samples. We compared Gradient Ascent Unlearning, Fisher Forgetting, and Full Retraining.\u003c/p\u003e \u003cp\u003eOur classifier achieved 85.58% validation accuracy, 79.34% balanced accuracy, and a macro-F1 score of 75.15%. Among unlearning methods, Gradient Ascent was fastest (11.01% forget-set accuracy in 103.6 seconds), while Full Retraining gave the best retention (99.64% at 2649.4 seconds). Our experiments also reveal an important implementation constraint: Fisher Forgetting can collapse when the numerical stabilizer is not large enough. Parameters with low importance receive substantial noise perturbations instead of selective forgetting. This is a non-trivial issue with significant practical implications.\u003c/p\u003e","manuscriptTitle":"ResNet34-Based Galaxy Morphology Classification with Machine Unlearning","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-04-07 03:31:33","doi":"10.21203/rs.3.rs-9269285/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":"b13280ca-74f5-4d7e-912b-63e6dc7e1509","owner":[],"postedDate":"April 7th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-04-12T16:10:10+00:00","versionOfRecord":[],"versionCreatedAt":"2026-04-07 03:31:33","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9269285","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9269285","identity":"rs-9269285","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}