Efficient two-stage embedded feature selection for bee sound classification

Efficient two-stage embedded feature selection for bee sound classification | 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 Efficient two-stage embedded feature selection for bee sound classification Nguyen Tran Dinh Ho, Dac Toan Ho, Thi-Thu-Hong Phan This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9312340/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 6 You are reading this latest preprint version Abstract Accurate classification of bee sounds is a fundamental task for acoustic-based monitoring systems. In this work, we propose a hybrid deep learning–machine learning (DL–ML) framework that integrates multi-architecture feature extraction with a two-stage embedded feature selection strategy. The proposed approach first transforms 1D audio signals into 2D Mel-frequency cepstral coefficient (MFCC) spectrograms, enabling deep neural networks to extract high-level acoustic representations. To effectively handle high-dimensional and redundant features, we introduce a two-stage feature selection process: the first stage applies feature selection independently to the deep features extracted from each model to eliminate intra-model redundancy, while the second stage performs feature selection on the fused feature space to remove inter-model redundancy. By explicitly addressing both sources of redundancy, the proposed framework produces compact yet highly discriminative feature representations. Experimental results on three datasets (Buzz 1, Buzz 2, and HN20K) demonstrate that the approach achieves accuracies of 98.87%, 99.61%, and 98.18%, respectively, while maintaining performance with up to 95%-99% feature reduction. The proposed two-stage feature selection framework provides an effective and computationally efficient solution for robust bee sound classification, with potential applicability to real-time acoustic monitoring systems. Bee sound classification Hybrid DL–ML framework MFCC Feature fusion Two-stage feature selection Embedded feature selection Full Text Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Reviews received at journal 24 Apr, 2026 Reviewers agreed at journal 19 Apr, 2026 Reviewers invited by journal 18 Apr, 2026 Editor assigned by journal 08 Apr, 2026 Submission checks completed at journal 08 Apr, 2026 First submitted to journal 03 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-9312340","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":628820324,"identity":"5e755a87-210b-421f-ac48-2c8ec55bdf85","order_by":0,"name":"Nguyen Tran Dinh Ho","email":"","orcid":"","institution":"FPT University","correspondingAuthor":false,"prefix":"","firstName":"Nguyen","middleName":"Tran Dinh","lastName":"Ho","suffix":""},{"id":628820325,"identity":"eaf1643a-396a-4ebc-9c5d-f5fa965f098b","order_by":1,"name":"Dac Toan Ho","email":"","orcid":"","institution":"FPT University","correspondingAuthor":false,"prefix":"","firstName":"Dac","middleName":"Toan","lastName":"Ho","suffix":""},{"id":628820326,"identity":"7cb166b2-f251-4fa8-948d-144011ad1312","order_by":2,"name":"Thi-Thu-Hong Phan","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAz0lEQVRIiWNgGAWjYBACAyjNw8be+ABIMxOrJcFAho/nsAFpWmzkJJKJ1GLO3nv4deGPPzxsko8ZHzBUWCc2SOQewKvFsudcmvWMBAMeNulkZgOGM+lALXkJ+B12I8fMmAesJf+YBGPb4cQGnjMGRGqRPMz+g/EfcVqMH4O1SDCzMTA2ALWw9xDQcuaMGfOMNGMeNp5kZomEY+nGbQS1HO8x/lxgI2cv336Y8cOHGmvZfmYe/FqAgE0azkwAcQmpBwLmz0QoGgWjYBSMgpEMAACiPIupLE5HAAAAAElFTkSuQmCC","orcid":"","institution":"FPT University","correspondingAuthor":true,"prefix":"","firstName":"Thi-Thu-Hong","middleName":"","lastName":"Phan","suffix":""}],"badges":[],"createdAt":"2026-04-03 11:44:28","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9312340/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9312340/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":107870795,"identity":"f56bdbba-40c2-4923-a8e3-66ba6b0bba66","added_by":"auto","created_at":"2026-04-27 07:40:44","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":7151994,"visible":true,"origin":"","legend":"","description":"","filename":"2026MedesJournalBeesoudselectionextend.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9312340/v1_covered_037bee85-d202-4ba2-87f1-5be2f8546598.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Efficient two-stage embedded feature selection for bee sound classification","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":"world-wide-web","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"wwwj","sideBox":"Learn more about [World Wide Web](http://link.springer.com/journal/11280)","snPcode":"11280","submissionUrl":"https://submission.nature.com/new-submission/11280/3","title":"World Wide Web","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"Bee sound classification, Hybrid DL–ML framework, MFCC, Feature fusion, Two-stage feature selection, Embedded feature selection","lastPublishedDoi":"10.21203/rs.3.rs-9312340/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9312340/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Accurate classification of bee sounds is a fundamental task for acoustic-based monitoring systems. In this work, we propose a hybrid deep learning–machine learning (DL–ML) framework that integrates multi-architecture feature extraction with a two-stage embedded feature selection strategy. The proposed approach first transforms 1D audio signals into 2D Mel-frequency cepstral coefficient (MFCC) spectrograms, enabling deep neural networks to extract high-level acoustic representations. To effectively handle high-dimensional and redundant features, we introduce a two-stage feature selection process: the first stage applies feature selection independently to the deep features extracted from each model to eliminate intra-model redundancy, while the second stage performs feature selection on the fused feature space to remove inter-model redundancy. By explicitly addressing both sources of redundancy, the proposed framework produces compact yet highly discriminative feature representations. Experimental results on three datasets (Buzz 1, Buzz 2, and HN20K) demonstrate that the approach achieves accuracies of 98.87%, 99.61%, and 98.18%, respectively, while maintaining performance with up to 95%-99% feature reduction. The proposed two-stage feature selection framework provides an effective and computationally efficient solution for robust bee sound classification, with potential applicability to real-time acoustic monitoring systems.","manuscriptTitle":"Efficient two-stage embedded feature selection for bee sound classification","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-04-27 06:00:35","doi":"10.21203/rs.3.rs-9312340/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"editorInvitedReview","content":"","date":"2026-04-24T07:58:28+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"312259341908292864868024439285316658034","date":"2026-04-20T02:38:49+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-04-18T14:54:29+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-04-08T04:29:31+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-04-08T04:17:34+00:00","index":"","fulltext":""},{"type":"submitted","content":"World Wide Web","date":"2026-04-03T11:08:39+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"world-wide-web","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"wwwj","sideBox":"Learn more about [World Wide Web](http://link.springer.com/journal/11280)","snPcode":"11280","submissionUrl":"https://submission.nature.com/new-submission/11280/3","title":"World Wide Web","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"71c572f9-73a3-4eb3-8c03-a4db8c4d2ced","owner":[],"postedDate":"April 27th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-04-27T06:00:35+00:00","versionOfRecord":[],"versionCreatedAt":"2026-04-27 06:00:35","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9312340","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9312340","identity":"rs-9312340","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}