XTinyHAR: A Tiny Inertial Transformer for Human Activity Recognition via Multimodal Knowledge Distillation and Explainable AI

XTinyHAR: A Tiny Inertial Transformer for Human Activity Recognition via Multimodal Knowledge Distillation and Explainable AI | 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 Article XTinyHAR: A Tiny Inertial Transformer for Human Activity Recognition via Multimodal Knowledge Distillation and Explainable AI Ismail Lamaakal, Chaymae Yahyati, yassine Maleh, Khalid El Makkaoui, and 4 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7284766/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 27 Nov, 2025 Read the published version in Scientific Reports → Version 1 posted 17 You are reading this latest preprint version Abstract Human Activity Recognition (HAR) is essential for applications such as healthcare monitoring, fitness tracking, and smart environments, yet deploying accurate and interpretable models on resource-constrained devices remains challenging. In this paper, we propose XTinyHAR, a lightweight, transformer-based unimodal framework trained via cross-modal knowledge distillation from a multimodal teacher. Our model incorporates temporal positional embeddings and attention rollout to enhance sequential feature extraction and interpretability. Evaluated on UTD-MHAD and MM-Fit datasets, XTinyHAR achieves state- of-the-art performance with test accuracies of 98.71% and 98.55%, F1-scores of 98.71% and 98.55%, and Cohen’s Kappa scores above 0.98, while maintaining a compact footprint of 2.45 MB, fast inference latency (3.1 ms CPU, 1.2 ms GPU), and low computational cost (11.3M FLOPs). Extensive ablation studies confirm the contribution of each component, and subject-wise evaluations demonstrate strong generalization across users. These results highlight XTinyHAR’s potential as a high-performance, interpretable, and deployable solution for real-time HAR on edge devices. Biological sciences/Computational biology and bioinformatics Physical sciences/Engineering Health sciences/Health care Physical sciences/Mathematics and computing Full Text Additional Declarations No competing interests reported. Cite Share Download PDF Status: Published Journal Publication published 27 Nov, 2025 Read the published version in Scientific Reports → Version 1 posted Editorial decision: Revision requested 12 Sep, 2025 Reviews received at journal 09 Sep, 2025 Reviews received at journal 06 Sep, 2025 Reviews received at journal 04 Sep, 2025 Reviewers agreed at journal 03 Sep, 2025 Reviews received at journal 02 Sep, 2025 Reviewers agreed at journal 29 Aug, 2025 Reviewers agreed at journal 27 Aug, 2025 Reviews received at journal 26 Aug, 2025 Reviewers agreed at journal 26 Aug, 2025 Reviewers agreed at journal 26 Aug, 2025 Reviewers agreed at journal 25 Aug, 2025 Reviewers invited by journal 25 Aug, 2025 Editor invited by journal 08 Aug, 2025 Editor assigned by journal 05 Aug, 2025 Submission checks completed at journal 04 Aug, 2025 First submitted to journal 03 Aug, 2025 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-7284766","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":507341776,"identity":"80aff381-c32e-44fe-baaf-92867b80847f","order_by":0,"name":"Ismail Lamaakal","email":"","orcid":"","institution":"Mohamed I University","correspondingAuthor":false,"prefix":"","firstName":"Ismail","middleName":"","lastName":"Lamaakal","suffix":""},{"id":507341777,"identity":"879b62f7-f5c9-44a9-b984-72c2fbf687f0","order_by":1,"name":"Chaymae Yahyati","email":"","orcid":"","institution":"Mohamed I University","correspondingAuthor":false,"prefix":"","firstName":"Chaymae","middleName":"","lastName":"Yahyati","suffix":""},{"id":507341778,"identity":"6490f2a7-5759-4716-8f7c-517d629032d8","order_by":2,"name":"yassine Maleh","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAsklEQVRIiWNgGAWjYLCCBwwH5EjUksBwwJh0LYkNRKs2OMCd+CGh4k76hvOnkz8w5tgRo4V3s0TCmWe5G27kbpNg3JZMlJYNEolth4FaeLcxMG5jJs6WH4n/DqcbnD+7+QPjtnqitGyTSGw4nGBwIHcD0GGHCWuRPMy7zSLh2GHDmSC/JG47TlgL3/HezTc+1ByW5wM57OO2asJaFFBckkBYAwODfAMxqkbBKBgFo2BkAwCmC0MSQYde8wAAAABJRU5ErkJggg==","orcid":"","institution":"Université Sultan Moulay Slimane","correspondingAuthor":true,"prefix":"","firstName":"yassine","middleName":"","lastName":"Maleh","suffix":""},{"id":507341779,"identity":"09926c9f-1fa4-4296-b31a-3f08dca941c8","order_by":3,"name":"Khalid El Makkaoui","email":"","orcid":"","institution":"Mohamed I University","correspondingAuthor":false,"prefix":"","firstName":"Khalid","middleName":"El","lastName":"Makkaoui","suffix":""},{"id":507341780,"identity":"f0d6460b-aadc-4f16-8f8f-21bda79e42cd","order_by":4,"name":"Ibrahim Ouahbi","email":"","orcid":"","institution":"Mohamed I University","correspondingAuthor":false,"prefix":"","firstName":"Ibrahim","middleName":"","lastName":"Ouahbi","suffix":""},{"id":507341781,"identity":"0b6f8e14-32ff-4d1e-9b67-560753553f50","order_by":5,"name":"Ahmed A. Abd El-Latif","email":"","orcid":"","institution":"Prince Sultan University","correspondingAuthor":false,"prefix":"","firstName":"Ahmed","middleName":"A. Abd","lastName":"El-Latif","suffix":""},{"id":507341782,"identity":"9917e67d-dab6-44ae-8f83-6fffe84d4124","order_by":6,"name":"Mariam Zomorodi","email":"","orcid":"","institution":"Cracow University of Technology","correspondingAuthor":false,"prefix":"","firstName":"Mariam","middleName":"","lastName":"Zomorodi","suffix":""},{"id":507341783,"identity":"cdd7400a-dfb5-4417-a880-5ad504da5832","order_by":7,"name":"Basma Abd El-Rahiem","email":"","orcid":"","institution":"Menoufia University","correspondingAuthor":false,"prefix":"","firstName":"Basma","middleName":"Abd","lastName":"El-Rahiem","suffix":""}],"badges":[],"createdAt":"2025-08-03 17:23:15","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7284766/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7284766/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s41598-025-26297-2","type":"published","date":"2025-11-27T15:57:44+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":97178729,"identity":"a1e027a9-31af-4ebb-816d-636b46779a4c","added_by":"auto","created_at":"2025-12-01 16:13:03","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":3983515,"visible":true,"origin":"","legend":"","description":"","filename":"ScientificReportIsmailPaper4PhDthesis.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7284766/v1_covered_30b2ed00-54d9-413d-9ced-e6ba3f0449b0.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"XTinyHAR: A Tiny Inertial Transformer for Human Activity Recognition via Multimodal Knowledge Distillation and Explainable AI","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-7284766/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7284766/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Human Activity Recognition (HAR) is essential for applications such as healthcare monitoring, fitness tracking, and smart environments, yet deploying accurate and interpretable models on resource-constrained devices remains challenging. In this paper, we propose XTinyHAR, a lightweight, transformer-based unimodal framework trained via cross-modal knowledge distillation from a multimodal teacher. Our model incorporates temporal positional embeddings and attention rollout to enhance sequential feature extraction and interpretability. Evaluated on UTD-MHAD and MM-Fit datasets, XTinyHAR achieves state- of-the-art performance with test accuracies of 98.71% and 98.55%, F1-scores of 98.71% and 98.55%, and Cohen’s Kappa scores above 0.98, while maintaining a compact footprint of 2.45 MB, fast inference latency (3.1 ms CPU, 1.2 ms GPU), and low computational cost (11.3M FLOPs). Extensive ablation studies confirm the contribution of each component, and subject-wise evaluations demonstrate strong generalization across users. These results highlight XTinyHAR’s potential as a high-performance, interpretable, and deployable solution for real-time HAR on edge devices.\n","manuscriptTitle":"XTinyHAR: A Tiny Inertial Transformer for Human Activity Recognition via Multimodal Knowledge Distillation and Explainable AI","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-09-02 02:44:05","doi":"10.21203/rs.3.rs-7284766/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-09-12T08:03:32+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-09-09T05:39:01+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-09-06T22:21:42+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-09-05T01:01:48+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"38202655591829666099726480072621250048","date":"2025-09-03T17:31:13+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-09-02T08:50:30+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"90286879574425519654831218085714815670","date":"2025-08-29T08:41:51+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"121340125242645816914810502533086770681","date":"2025-08-27T12:28:08+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-08-27T01:40:57+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"335874092235005980933014528775038547962","date":"2025-08-26T13:54:53+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"16779207467607641898173155962812459287","date":"2025-08-26T12:04:14+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"301085879124988789445151838256621368128","date":"2025-08-26T02:02:48+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-08-25T12:13:37+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2025-08-08T06:57:33+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-08-05T06:19:15+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-08-05T03:17:18+00:00","index":"","fulltext":""},{"type":"submitted","content":"Scientific Reports","date":"2025-08-03T17:17:56+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"f036f324-9bc0-4c4d-a4a1-69ceaed843ed","owner":[],"postedDate":"September 2nd, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":53888971,"name":"Biological sciences/Computational biology and bioinformatics"},{"id":53888972,"name":"Physical sciences/Engineering"},{"id":53888973,"name":"Health sciences/Health care"},{"id":53888974,"name":"Physical sciences/Mathematics and computing"}],"tags":[],"updatedAt":"2025-12-01T16:06:29+00:00","versionOfRecord":{"articleIdentity":"rs-7284766","link":"https://doi.org/10.1038/s41598-025-26297-2","journal":{"identity":"scientific-reports","isVorOnly":false,"title":"Scientific Reports"},"publishedOn":"2025-11-27 15:57:44","publishedOnDateReadable":"November 27th, 2025"},"versionCreatedAt":"2025-09-02 02:44:05","video":"","vorDoi":"10.1038/s41598-025-26297-2","vorDoiUrl":"https://doi.org/10.1038/s41598-025-26297-2","workflowStages":[]},"version":"v1","identity":"rs-7284766","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7284766","identity":"rs-7284766","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}