A Novel High-Speed Optical Computing Platform with Dynamic In-situ Reconfigurability

A Novel High-Speed Optical Computing Platform with Dynamic In-situ Reconfigurability | 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 A Novel High-Speed Optical Computing Platform with Dynamic In-situ Reconfigurability Kenneth Kin-Yip Wong, Yuanjia Wang, Xin Dong, Pujin Cheng, Yi Zhou This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9674340/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted You are reading this latest preprint version Abstract Artificial intelligence (AI) on resource-constrained edge platforms is increasingly encountering computational bottlenecks as electronic hardware approaches the physical limits of Moore’s Law. Optical neural networks (ONNs) present a promising alternative computational paradigm. However, the practical deployment of current ONNs is restricted by challenges related to system speed, in-situ reconfigurability, and overall hardware complexity. Hence, this work proposes HS-BONN: a high-speed binarized ONN built upon a novel multi-reflection and spatial-partitioning digital micromirror device (DMD) platform. This architecture enables rapid optical forward propagation, dynamic in-situ reconfigurability, and highly efficient resource utilization. By synergizing the discrete logic of Binarized Neural Networks (BNNs) with the DMD, we realize a complete multi-layer 2D fully-connected neural network. HS-BONN achieves a peak processing speed of 460 TOPS while maintaining accuracy superior to equivalent electronic BNNs. Leveraging its rapid in-situ reconfigurability, we demonstrate an ONN-driven multi-task scheduler tailored for advanced GPT-style interfaces. Evaluated on a complex 4-type mixed dataset, the system attains an overall accuracy of 92.27%, accompanied by a 96.88% reduction in memory footprint and a remarkably low power consumption of just 2.5 W. Furthermore, the platform also exhibits enhanced robustness against the commonly encountered environmental noises in real-world deployments. Physical sciences/Mathematics and computing/Computational science Physical sciences/Optics and photonics/Applied optics/Optoelectronic devices and components Physical sciences/Optics and photonics/Optical physics/Ultrafast photonics Optical neural network high-speed computing platform reconfigurable multi-layer network binarized neural network GPT-oriented applications Full Text Additional Declarations There is NO Competing Interest. Supplementary Files SupplementaryHSBONN.pdf Supplementary for HS_BONN manuscript Cite Share Download PDF Status: Under Review 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-9674340","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":638288787,"identity":"fe0743b2-5954-431f-a2a7-e5d73e4f33b3","order_by":0,"name":"Kenneth Kin-Yip Wong","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAz0lEQVRIiWNgGAWjYBACCQhlAyLYGBsYGBIkiNSSRrqWwyRokZyRY/i44Nd5OX72BraHM9sY8iQbCGiRlsgxNp7Zd9tYsucAu+HGNoZiaUK2yEnkbpPm7bmduOFGApvkwzaGxHlEajlXv//+AyK1SIO08Pw4kGAgwcAmCXRY4mxCWiR73n825m1INpxxJrFNcsY5icSZDQS0SBxPS3zM88dOnr/98DHJnjKbxBkHCFkjkMDAwNgGYoFihYFgrAABP8jQP0QoHAWjYBSMgpELAPMCP2PbsiALAAAAAElFTkSuQmCC","orcid":"https://orcid.org/0000-0003-3668-3539","institution":"University of Hong Kong","correspondingAuthor":true,"prefix":"","firstName":"Kenneth","middleName":"Kin-Yip","lastName":"Wong","suffix":""},{"id":638288788,"identity":"b9fa4eca-cb08-4745-8f83-a8710e1a17b7","order_by":1,"name":"Yuanjia Wang","email":"","orcid":"","institution":"University of Hong Kong","correspondingAuthor":false,"prefix":"","firstName":"Yuanjia","middleName":"","lastName":"Wang","suffix":""},{"id":638288789,"identity":"9a162781-ebd3-408a-ae4d-8c398e832d57","order_by":2,"name":"Xin Dong","email":"","orcid":"","institution":"The University of Hong Kong","correspondingAuthor":false,"prefix":"","firstName":"Xin","middleName":"","lastName":"Dong","suffix":""},{"id":638288790,"identity":"401bb7c3-8c3b-4cb1-b571-215f926a3360","order_by":3,"name":"Pujin Cheng","email":"","orcid":"","institution":"Department of Electrical and Electronic Engineering, Southern University of Science and Technology.","correspondingAuthor":false,"prefix":"","firstName":"Pujin","middleName":"","lastName":"Cheng","suffix":""},{"id":638288791,"identity":"ab085f48-16ee-46b7-a8a7-07fa9244ab67","order_by":4,"name":"Yi Zhou","email":"","orcid":"","institution":"National University of Defense Technology","correspondingAuthor":false,"prefix":"","firstName":"Yi","middleName":"","lastName":"Zhou","suffix":""}],"badges":[],"createdAt":"2026-05-11 04:11:48","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9674340/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9674340/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":109096526,"identity":"b7584fd1-4421-49fa-a4ec-aa959087f22c","added_by":"auto","created_at":"2026-05-12 14:03:35","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":6211132,"visible":true,"origin":"","legend":"Article File","description":"","filename":"HSBONNmanuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9674340/v1_covered_1f47e802-daf9-4b95-842d-d1302c531d2a.pdf"},{"id":109095390,"identity":"d00c749a-e33f-4601-86cc-ef2b8e392a5a","added_by":"auto","created_at":"2026-05-12 13:57:26","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":3770200,"visible":true,"origin":"","legend":"Supplementary for HS_BONN manuscript","description":"","filename":"SupplementaryHSBONN.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9674340/v1/75837acfc9305d70783baf4e.pdf"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"A Novel High-Speed Optical Computing Platform with Dynamic In-situ Reconfigurability","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"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":"nature-portfolio","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"Nature Portfolio","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"ejp","reportingPortfolio":"","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"Optical neural network, high-speed computing platform, reconfigurable multi-layer network, binarized neural network, GPT-oriented applications","lastPublishedDoi":"10.21203/rs.3.rs-9674340/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9674340/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Artificial intelligence (AI) on resource-constrained edge platforms is increasingly encountering computational bottlenecks as electronic hardware approaches the physical limits of Moore’s Law. Optical neural networks (ONNs) present a promising alternative computational paradigm. However, the practical deployment of current ONNs is restricted by challenges related to system speed, in-situ reconfigurability, and overall hardware complexity. Hence, this work proposes HS-BONN: a high-speed binarized ONN built upon a novel multi-reflection and spatial-partitioning digital micromirror device (DMD) platform. This architecture enables rapid optical forward propagation, dynamic in-situ reconfigurability, and highly efficient resource utilization. By synergizing the discrete logic of Binarized Neural Networks (BNNs) with the DMD, we realize a complete multi-layer 2D fully-connected neural network. HS-BONN achieves a peak processing speed of 460 TOPS while maintaining accuracy superior to equivalent electronic BNNs. Leveraging its rapid in-situ reconfigurability, we demonstrate an ONN-driven multi-task scheduler tailored for advanced GPT-style interfaces. Evaluated on a complex 4-type mixed dataset, the system attains an overall accuracy of 92.27%, accompanied by a 96.88% reduction in memory footprint and a remarkably low power consumption of just 2.5 W. Furthermore, the platform also exhibits enhanced robustness against the commonly encountered environmental noises in real-world deployments.","manuscriptTitle":"A Novel High-Speed Optical Computing Platform with Dynamic In-situ Reconfigurability","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-05-12 13:30:41","doi":"10.21203/rs.3.rs-9674340/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"nature-communications","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"NCOMMS","sideBox":"Learn more about [Nature Communications](http://www.nature.com/ncomms/)","snPcode":"","submissionUrl":"https://mts-ncomms.nature.com/","title":"Nature Communications","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Nature Communications","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"45251999-c63d-4d96-b54f-1a1655d859eb","owner":[],"postedDate":"May 12th, 2026","published":true,"recentEditorialEvents":[{"type":"reviewerAgreed","content":"This content is not available.","date":"2026-05-12T01:25:34+00:00","index":1,"fulltext":"This content is not available."},{"type":"reviewersInvited","content":"6","date":"2026-05-11T13:17:02+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-05-11T09:26:50+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-05-11T08:44:38+00:00","index":"","fulltext":""},{"type":"submitted","content":"Nature Communications","date":"2026-05-11T04:09:36+00:00","index":"","fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":67934479,"name":"Physical sciences/Mathematics and computing/Computational science"},{"id":67934480,"name":"Physical sciences/Optics and photonics/Applied optics/Optoelectronic devices and components"},{"id":67934481,"name":"Physical sciences/Optics and photonics/Optical physics/Ultrafast photonics"}],"tags":[],"updatedAt":"2026-05-12T13:30:41+00:00","versionOfRecord":[],"versionCreatedAt":"2026-05-12 13:30:41","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9674340","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9674340","identity":"rs-9674340","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}