{"paper_id":"115e269f-fd88-4d02-8ef0-47cc6d89252a","body_text":"LoRAE: Low-Rank Adaptation for Edge 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PDF Article LoRAE: Low-Rank Adaptation for Edge AI Zhixue Wang, Hongyao Ma, Jiahui Zhai This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6750974/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 26 Sep, 2025 Read the published version in Scientific Reports → Version 1 posted 11 You are reading this latest preprint version Abstract The rapid advancement of edge artificial intelligence (AI) has unlocked transformative applications across various domains. However, it also poses significant challenges in efficiently updating models on edge devices, which are often constrained by limited computational and communication resources. Here, we present low-rank adaptation method for Edge AI (LoRAE), Leveraging low-rank decomposition of convolutional neural networks (CNNs) weight matrices, LoRAE reduces the number of updated parameters to approximately 4% of traditional full-parameter updates, effectively mitigating the computational and communication challenges associated with model updates. Extensive experiments across image classification, object detection, and image segmentation tasks demonstrate that LoRAE significantly decreases the scale of trainable parameters while maintaining or even enhancing model accuracy. Using the YOLOv8x model, LoRAE achieves parameter reductions of 86.1%, 98.6%, and 94.1% across the three tasks, respectively, without compromising accuracy. These findings highlight the potential of LoRAE as an efficient and precise solution for resource-constrained edge AI systems. Physical sciences/Mathematics and computing/Computer science Physical sciences/Mathematics and computing/Computational science Full Text Additional Declarations No competing interests reported. Cite Share Download PDF Status: Published Journal Publication published 26 Sep, 2025 Read the published version in Scientific Reports → Version 1 posted Editorial decision: Revision requested 07 Jul, 2025 Reviews received at journal 25 Jun, 2025 Reviewers agreed at journal 25 Jun, 2025 Reviews received at journal 14 Jun, 2025 Reviewers agreed at journal 07 Jun, 2025 Reviewers agreed at journal 03 Jun, 2025 Reviewers invited by journal 03 Jun, 2025 Editor invited by journal 31 May, 2025 Editor assigned by journal 28 May, 2025 Submission checks completed at journal 27 May, 2025 First submitted to journal 26 May, 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-6750974\",\"acceptedTermsAndConditions\":true,\"allowDirectSubmit\":false,\"archivedVersions\":[],\"articleType\":\"Article\",\"associatedPublications\":[],\"authors\":[{\"id\":465945191,\"identity\":\"115cb731-d140-440f-8548-684b4bd95aea\",\"order_by\":0,\"name\":\"Zhixue Wang\",\"email\":\"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAvklEQVRIiWNgGAWjYDACCcYGMM3PzHz4AWlaJNvZ0gyI1AKlDc7zKEjgVQkD8rOb26R5KhgSNx/mYTBgqLGJJqiFcc7BZmOeMwyJ2w7zHnjAcCwtt4GQFmaJxMbHvG0MudsO8yUYMDYcJqyFTSKx4TDvP4bczc08BhJEaeEB29LAkLuBmVgtEhKJzYZzjknUzzgMDOQEYvwiPyP9mcSbGhtj/v7Dhx98qLEhrAVmGYRKIFL5KBgFo2AUjAICAAAV1DmAaBVM3gAAAABJRU5ErkJggg==\",\"orcid\":\"\",\"institution\":\"Shandong Jiaotong University\",\"correspondingAuthor\":true,\"prefix\":\"\",\"firstName\":\"Zhixue\",\"middleName\":\"\",\"lastName\":\"Wang\",\"suffix\":\"\"},{\"id\":465945192,\"identity\":\"e7f8216a-fed6-4b55-8bac-3c9cb52929f7\",\"order_by\":1,\"name\":\"Hongyao Ma\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"Beijing University of Technology\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Hongyao\",\"middleName\":\"\",\"lastName\":\"Ma\",\"suffix\":\"\"},{\"id\":465945193,\"identity\":\"5b222471-92bf-4c5a-a859-3d96948b057d\",\"order_by\":2,\"name\":\"Jiahui Zhai\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"Beijing University of Technology\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Jiahui\",\"middleName\":\"\",\"lastName\":\"Zhai\",\"suffix\":\"\"}],\"badges\":[],\"createdAt\":\"2025-05-26 12:53:10\",\"currentVersionCode\":1,\"declarations\":\"\",\"doi\":\"10.21203/rs.3.rs-6750974/v1\",\"doiUrl\":\"https://doi.org/10.21203/rs.3.rs-6750974/v1\",\"draftVersion\":[],\"editorialEvents\":[{\"content\":\"https://doi.org/10.1038/s41598-025-16794-9\",\"type\":\"published\",\"date\":\"2025-09-26T15:57:51+00:00\"}],\"editorialNote\":\"\",\"failedWorkflow\":false,\"files\":[{\"id\":92430558,\"identity\":\"ad4c8c12-bceb-4688-9f83-5e540e3f7799\",\"added_by\":\"auto\",\"created_at\":\"2025-09-29 16:05:43\",\"extension\":\"pdf\",\"order_by\":1,\"title\":\"\",\"display\":\"\",\"copyAsset\":false,\"role\":\"manuscript-pdf\",\"size\":2345473,\"visible\":true,\"origin\":\"\",\"legend\":\"\",\"description\":\"\",\"filename\":\"LoRAELowRankAdaptationforEdgeAIScientificreports2.pdf\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-6750974/v1_covered_88709a3c-1cb4-4534-9bfc-7a02d0468721.pdf\"}],\"financialInterests\":\"No competing interests reported.\",\"formattedTitle\":\"LoRAE: Low-Rank Adaptation for Edge 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\":\"info@researchsquare.com\",\"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-6750974/v1\",\"lastPublishedDoiUrl\":\"https://doi.org/10.21203/rs.3.rs-6750974/v1\",\"license\":{\"name\":\"CC BY 4.0\",\"url\":\"https://creativecommons.org/licenses/by/4.0/\"},\"manuscriptAbstract\":\"The rapid advancement of edge artificial intelligence (AI) has unlocked transformative applications across various domains. However, it also poses significant challenges in efficiently updating models on edge devices, which are often constrained by limited computational and communication resources. Here, we present low-rank adaptation method for Edge AI (LoRAE), Leveraging low-rank decomposition of convolutional neural networks (CNNs) weight matrices, LoRAE reduces the number of updated parameters to approximately 4% of traditional full-parameter updates, effectively mitigating the computational and communication challenges associated with model updates. Extensive experiments across image classification, object detection, and image segmentation tasks demonstrate that LoRAE significantly decreases the scale of trainable parameters while maintaining or even enhancing model accuracy. Using the YOLOv8x model, LoRAE achieves parameter reductions of 86.1%, 98.6%, and 94.1% across the three tasks, respectively, without compromising accuracy. These findings highlight the potential of LoRAE as an efficient and precise solution for resource-constrained edge AI systems.\",\"manuscriptTitle\":\"LoRAE: Low-Rank Adaptation for Edge AI\",\"msid\":\"\",\"msnumber\":\"\",\"nonDraftVersions\":[{\"code\":1,\"date\":\"2025-06-06 06:43:16\",\"doi\":\"10.21203/rs.3.rs-6750974/v1\",\"editorialEvents\":[{\"type\":\"communityComments\",\"content\":0},{\"type\":\"decision\",\"content\":\"Revision requested\",\"date\":\"2025-07-07T11:51:53+00:00\",\"index\":\"\",\"fulltext\":\"\"},{\"type\":\"editorInvitedReview\",\"content\":\"\",\"date\":\"2025-06-26T03:49:34+00:00\",\"index\":\"hide\",\"fulltext\":\"\"},{\"type\":\"reviewerAgreed\",\"content\":\"275403210174693943655597158209371520537\",\"date\":\"2025-06-25T05:10:18+00:00\",\"index\":\"hide\",\"fulltext\":\"\"},{\"type\":\"editorInvitedReview\",\"content\":\"\",\"date\":\"2025-06-14T09:56:54+00:00\",\"index\":\"hide\",\"fulltext\":\"\"},{\"type\":\"reviewerAgreed\",\"content\":\"129951499811907375880765856602893881019\",\"date\":\"2025-06-07T09:25:47+00:00\",\"index\":\"hide\",\"fulltext\":\"\"},{\"type\":\"reviewerAgreed\",\"content\":\"259290810050492993163886984572331200443\",\"date\":\"2025-06-03T14:49:42+00:00\",\"index\":\"hide\",\"fulltext\":\"\"},{\"type\":\"reviewersInvited\",\"content\":\"\",\"date\":\"2025-06-03T13:56:24+00:00\",\"index\":\"\",\"fulltext\":\"\"},{\"type\":\"editorInvited\",\"content\":\"\",\"date\":\"2025-05-31T14:50:07+00:00\",\"index\":\"\",\"fulltext\":\"\"},{\"type\":\"editorAssigned\",\"content\":\"\",\"date\":\"2025-05-28T12:29:09+00:00\",\"index\":\"\",\"fulltext\":\"\"},{\"type\":\"checksComplete\",\"content\":\"\",\"date\":\"2025-05-28T01:18:06+00:00\",\"index\":\"\",\"fulltext\":\"\"},{\"type\":\"submitted\",\"content\":\"Scientific Reports\",\"date\":\"2025-05-26T12:39:38+00:00\",\"index\":\"\",\"fulltext\":\"\"}],\"status\":\"published\",\"journal\":{\"display\":true,\"email\":\"info@researchsquare.com\",\"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\":\"c1bc07f7-47ab-4c5c-9cca-cade9f679464\",\"owner\":[],\"postedDate\":\"June 6th, 2025\",\"published\":true,\"recentEditorialEvents\":[],\"rejectedJournal\":[],\"revision\":\"\",\"amendment\":\"\",\"status\":\"published-in-journal\",\"subjectAreas\":[{\"id\":49467106,\"name\":\"Physical sciences/Mathematics and computing/Computer science\"},{\"id\":49467107,\"name\":\"Physical sciences/Mathematics and computing/Computational science\"}],\"tags\":[],\"updatedAt\":\"2025-09-29T16:01:17+00:00\",\"versionOfRecord\":{\"articleIdentity\":\"rs-6750974\",\"link\":\"https://doi.org/10.1038/s41598-025-16794-9\",\"journal\":{\"identity\":\"scientific-reports\",\"isVorOnly\":false,\"title\":\"Scientific Reports\"},\"publishedOn\":\"2025-09-26 15:57:51\",\"publishedOnDateReadable\":\"September 26th, 2025\"},\"versionCreatedAt\":\"2025-06-06 06:43:16\",\"video\":\"\",\"vorDoi\":\"10.1038/s41598-025-16794-9\",\"vorDoiUrl\":\"https://doi.org/10.1038/s41598-025-16794-9\",\"workflowStages\":[]},\"version\":\"v1\",\"identity\":\"rs-6750974\",\"journalConfig\":\"researchsquare\"},\"__N_SSP\":true},\"page\":\"/article/[identity]/[[...version]]\",\"query\":{\"redirect\":\"/article/rs-6750974\",\"identity\":\"rs-6750974\",\"version\":[\"v1\"]},\"buildId\":\"8U1c8b4HqxoKbykW_rLl7\",\"isFallback\":false,\"isExperimentalCompile\":false,\"dynamicIds\":[84888],\"gssp\":true,\"scriptLoader\":[]}","source_license":"CC-BY-4.0","license_restricted":false}