An Uncertainty-Aware and Explainable Deep Learning Model for Facial Beauty Prediction

An Uncertainty-Aware and Explainable Deep Learning Model for Facial Beauty Prediction | 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 An Uncertainty-Aware and Explainable Deep Learning Model for Facial Beauty Prediction Djamel Eddine Boukhari, Ali Chemsa This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6941023/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 The automated prediction of facial beauty is a challenging computer vision task due to the inherent subjectivity and complexity of human perception. Traditional models typically predict a single, deterministic score, failing to capture the ambiguity and variance in human judgments. This paper introduces a novel probabilistic approach that frames facial beauty prediction as an uncertainty estimation problem. We propose a dual-head deep convolutional neural network (CNN) that predicts not only the mean beauty score but also the underlying uncertainty (variance) of its prediction. This is achieved using a state-of-the-art EfficientNetV2-S backbone and a custom Gaussian Negative Log-Likelihood (NLL) loss function, which encourages the model to learn its own confidence. By modeling predictions as a probability distribution, our approach provides a richer, more informative output. Furthermore, we leverage Gradient-weighted Class Activation Mapping (Grad-CAM) to provide visual explanations, highlighting the facial regions most influential to the model's predictions. Experiments conducted on the FBP5500 dataset across both 5-fold cross-validation and a fixed 60%-40% train-test split. Our model achieves new state-of-the-art performance, with a mean Pearson Correlation of 0.9277, surpassing existing single-model and complex ensemble methods. The results demonstrate that our probabilistic and explainable model achieves strong quantitative performance while simultaneously offering a valuable measure of prediction uncertainty, paving the way for more robust and interpretable models in subjective assessment tasks. The code is available at https://github.com/DjameleddineBoukhari/XAI-FBP Facial Beauty Prediction Computer Vision Deep Learning Uncertainty Quantification Explainable AI Regression Convolutional Neural Networks 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-6941023","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":481056492,"identity":"c6b26e01-9108-4a50-852c-f023abcad518","order_by":0,"name":"Djamel Eddine Boukhari","email":"data:image/png;base64,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","orcid":"","institution":"CRSTRA","correspondingAuthor":true,"prefix":"","firstName":"Djamel","middleName":"Eddine","lastName":"Boukhari","suffix":""},{"id":481056493,"identity":"618c2611-283f-422b-aa56-b9f68d292221","order_by":1,"name":"Ali Chemsa","email":"","orcid":"","institution":"University of El Oued","correspondingAuthor":false,"prefix":"","firstName":"Ali","middleName":"","lastName":"Chemsa","suffix":""}],"badges":[],"createdAt":"2025-06-20 18:23:13","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6941023/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6941023/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":86289400,"identity":"487edb57-c593-4eff-93c3-90ccdbc51f2f","added_by":"auto","created_at":"2025-07-09 02:49:08","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":582380,"visible":true,"origin":"","legend":"","description":"","filename":"mysubmission.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6941023/v1_covered_d56337ff-19ab-45bf-bed1-5066d27b7b72.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"An Uncertainty-Aware and Explainable Deep Learning Model for Facial Beauty Prediction","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":"Facial Beauty Prediction, Computer Vision, Deep Learning, Uncertainty Quantification, Explainable AI, Regression, Convolutional Neural Networks","lastPublishedDoi":"10.21203/rs.3.rs-6941023/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6941023/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThe automated prediction of facial beauty is a challenging computer vision task due to the inherent subjectivity and complexity of human perception. Traditional models typically predict a single, deterministic score, failing to capture the ambiguity and variance in human judgments. This paper introduces a novel probabilistic approach that frames facial beauty prediction as an uncertainty estimation problem. We propose a dual-head deep convolutional neural network (CNN) that predicts not only the mean beauty score but also the underlying uncertainty (variance) of its prediction. This is achieved using a state-of-the-art EfficientNetV2-S backbone and a custom Gaussian Negative Log-Likelihood (NLL) loss function, which encourages the model to learn its own confidence. By modeling predictions as a probability distribution, our approach provides a richer, more informative output. Furthermore, we leverage Gradient-weighted Class Activation Mapping (Grad-CAM) to provide visual explanations, highlighting the facial regions most influential to the model's predictions. Experiments conducted on the FBP5500 dataset across both 5-fold cross-validation and a fixed 60%-40% train-test split. Our model achieves new state-of-the-art performance, with a mean Pearson Correlation of 0.9277, surpassing existing single-model and complex ensemble methods. The results demonstrate that our probabilistic and explainable model achieves strong quantitative performance while simultaneously offering a valuable measure of prediction uncertainty, paving the way for more robust and interpretable models in subjective assessment tasks. The code is available at \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/DjameleddineBoukhari/XAI-FBP\u003c/span\u003e\u003cspan address=\"https://github.com/DjameleddineBoukhari/XAI-FBP\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e","manuscriptTitle":"An Uncertainty-Aware and Explainable Deep Learning Model for Facial Beauty Prediction","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-07-09 02:41:01","doi":"10.21203/rs.3.rs-6941023/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":"8e12c9a8-7ecc-4df6-9916-22ae8a4d4867","owner":[],"postedDate":"July 9th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-07-16T15:32:09+00:00","versionOfRecord":[],"versionCreatedAt":"2025-07-09 02:41:01","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6941023","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6941023","identity":"rs-6941023","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}