Modified Spike Backpropagation Design towards Highly Parallelable Hardware Implementation

Modified Spike Backpropagation Design towards Highly Parallelable Hardware Implementation | 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 Modified Spike Backpropagation Design towards Highly Parallelable Hardware Implementation Dayou Zhang, Yue Zhou, Vivian Zhao, Bin Gao, Jiawei Fu, Yibai Xue, and 5 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7318696/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 9 You are reading this latest preprint version Abstract This work presents a hardware-algorithm co-designed framework for neuromor-phic computing, enabling efficient supervised learning in spike-based neural architectures. First, synaptic updates are reformulated as low-rank outer products of forward spike vectors and backward error gradients via singular value decomposition (SVD), enabling direct parallelization on 1T1R arrays. Second, a stochastic computing scheme replaces conventional sequential updates with probabilistic pulse-driven modulation, achieving one-step full-matrix synaptic updates. Third, gradient stabilization techniques mitigate training instability in deep SNNs by addressing silent neuron and gradient explosion issues. 1 Evaluated on the ASL-DVS dynamic gesture recognition task, the framework maintains 84.7% accuracy with hardware-realistic 1T1R characteristics, while drastically reducing hardware update steps. This demonstrates a syner-gistic hardware-algorithm co-design where SVD-based approximation enables parallelization, stochastic computing achieves one-step updates, and gradient stabilization ensures trainability, advancing practical neuromorphic intelligence for edge sensing systems. Physical sciences/Engineering Physical sciences/Mathematics and computing Biological sciences/Neuroscience Spiking neural networks Spike-timing-dependent learning Neuromorphic hardware Stochastic computing Event-driven processing Full Text Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Editorial decision: Revision requested 23 Aug, 2025 Reviews received at journal 23 Aug, 2025 Reviews received at journal 16 Aug, 2025 Reviewers agreed at journal 14 Aug, 2025 Reviewers agreed at journal 12 Aug, 2025 Reviewers invited by journal 12 Aug, 2025 Editor assigned by journal 11 Aug, 2025 Submission checks completed at journal 08 Aug, 2025 First submitted to journal 07 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-7318696","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":501380759,"identity":"ae457cbe-f6de-425b-8189-977f028d7033","order_by":0,"name":"Dayou Zhang","email":"","orcid":"","institution":"Wuhan National Laboratory for Optoelectronics","correspondingAuthor":false,"prefix":"","firstName":"Dayou","middleName":"","lastName":"Zhang","suffix":""},{"id":501380760,"identity":"7daf1f4d-272c-45ff-88f8-bdec0d8911c7","order_by":1,"name":"Yue Zhou","email":"","orcid":"","institution":"The Hong Kong Polytechnic University","correspondingAuthor":false,"prefix":"","firstName":"Yue","middleName":"","lastName":"Zhou","suffix":""},{"id":501380761,"identity":"7cfa7ccf-550d-4604-a8d3-462317b594f0","order_by":2,"name":"Vivian Zhao","email":"","orcid":"","institution":"Huazhong University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Vivian","middleName":"","lastName":"Zhao","suffix":""},{"id":501380762,"identity":"f4baa1e6-e2f7-4f0e-9a96-a138f61b5c7d","order_by":3,"name":"Bin Gao","email":"","orcid":"","institution":"Tsinghua University","correspondingAuthor":false,"prefix":"","firstName":"Bin","middleName":"","lastName":"Gao","suffix":""},{"id":501380763,"identity":"c27da3bf-2b6e-497c-b88c-8c2273ddf3ed","order_by":4,"name":"Jiawei Fu","email":"","orcid":"","institution":"Huazhong University of Science and Technology and Hubei Yangtze Memory Laboratories","correspondingAuthor":false,"prefix":"","firstName":"Jiawei","middleName":"","lastName":"Fu","suffix":""},{"id":501380764,"identity":"ddb3b9b6-07fc-43cf-a9a3-0a80770cf23c","order_by":5,"name":"Yibai Xue","email":"","orcid":"","institution":"Huazhong University of Science and Technology and Hubei Yangtze Memory Laboratories","correspondingAuthor":false,"prefix":"","firstName":"Yibai","middleName":"","lastName":"Xue","suffix":""},{"id":501380765,"identity":"ebccaf07-86bb-41f8-872c-c9f0253d723f","order_by":6,"name":"Zhe Yang","email":"","orcid":"","institution":"Huazhong University of Science and Technology and Hubei Yangtze Memory Laboratories","correspondingAuthor":false,"prefix":"","firstName":"Zhe","middleName":"","lastName":"Yang","suffix":""},{"id":501380766,"identity":"078703f0-0de0-492a-8098-9d7bdb5bda2d","order_by":7,"name":"Yi Li","email":"","orcid":"","institution":"Huazhong University of Science and Technology and Hubei Yangtze Memory Laboratories","correspondingAuthor":false,"prefix":"","firstName":"Yi","middleName":"","lastName":"Li","suffix":""},{"id":501380767,"identity":"ec55cc85-feed-4421-834b-c24d93b1e542","order_by":8,"name":"Hao Tong","email":"","orcid":"","institution":"Huazhong University of Science and Technology and Hubei Yangtze Memory Laboratories","correspondingAuthor":false,"prefix":"","firstName":"Hao","middleName":"","lastName":"Tong","suffix":""},{"id":501380768,"identity":"decc1f3b-dfae-4bac-bc7a-dd11a861c417","order_by":9,"name":"Xiangshui Miao","email":"","orcid":"","institution":"Wuhan National Laboratory for Optoelectronics","correspondingAuthor":false,"prefix":"","firstName":"Xiangshui","middleName":"","lastName":"Miao","suffix":""},{"id":501380769,"identity":"000c31c7-57f2-4009-b477-76b2270e8d86","order_by":10,"name":"Yuhui He","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA2ElEQVRIiWNgGAWjYBACAyBmZmCQkDOA8JmJ12IMZDA2kKKFIXED0VrM2XvMpAvbLNK3S6Q/f8BQYZ3YwH72AF4tlj1nzKRntknk7pyRY9jAcCY9sYEnLwG/w27kmEnzArVsuJHD2MDYdjixQYLHAL+W+2/AWtINbqQ/bGD8R4yWGzxgLQkGNxIMGxgbiNFyJq3YmuechOGGM28MZyQcSzdu48khoOX44Y23ecrq5A2Opz/48KHGWraf/Qx+LQwMHEgKEoCYjYB6IGB/QFjNKBgFo2AUjGwAAJCMQwcxlJquAAAAAElFTkSuQmCC","orcid":"","institution":"Huazhong University of Science and Technology and Hubei Yangtze Memory Laboratories","correspondingAuthor":true,"prefix":"","firstName":"Yuhui","middleName":"","lastName":"He","suffix":""}],"badges":[],"createdAt":"2025-08-07 12:23:17","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7318696/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7318696/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":89432889,"identity":"a1f2773b-efaf-4251-920f-95ed4c7aa492","added_by":"auto","created_at":"2025-08-20 00:56:50","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1373222,"visible":true,"origin":"","legend":"","description":"","filename":"snarticle.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7318696/v1_covered_40635860-c472-427e-8b93-72eb923e1b4c.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Modified Spike Backpropagation Design towards Highly Parallelable Hardware Implementation","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":"npj-unconventional-computing","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"Learn more about [npj Unconventional Computing](https://www.nature.com/npjunconvcomput)","snPcode":"44335","submissionUrl":"https://submission.springernature.com/new-submission/44335/3","title":"npj Unconventional Computing","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"NPJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":" Spiking neural networks, Spike-timing-dependent learning, Neuromorphic hardware, Stochastic computing, Event-driven processing","lastPublishedDoi":"10.21203/rs.3.rs-7318696/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7318696/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"This work presents a hardware-algorithm co-designed framework for neuromor-phic computing, enabling efficient supervised learning in spike-based neural architectures. First, synaptic updates are reformulated as low-rank outer products of forward spike vectors and backward error gradients via singular value decomposition (SVD), enabling direct parallelization on 1T1R arrays. Second, a stochastic computing scheme replaces conventional sequential updates with probabilistic pulse-driven modulation, achieving one-step full-matrix synaptic updates. Third, gradient stabilization techniques mitigate training instability in deep SNNs by addressing silent neuron and gradient explosion issues. 1 Evaluated on the ASL-DVS dynamic gesture recognition task, the framework maintains 84.7% accuracy with hardware-realistic 1T1R characteristics, while drastically reducing hardware update steps. This demonstrates a syner-gistic hardware-algorithm co-design where SVD-based approximation enables parallelization, stochastic computing achieves one-step updates, and gradient stabilization ensures trainability, advancing practical neuromorphic intelligence for edge sensing systems.","manuscriptTitle":"Modified Spike Backpropagation Design towards Highly Parallelable Hardware Implementation","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-08-20 00:48:43","doi":"10.21203/rs.3.rs-7318696/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-08-24T01:19:31+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-08-24T00:24:29+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-08-17T01:30:30+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"146343506987734875955710316058532280390","date":"2025-08-14T23:48:46+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"168110979747946663333991583621503783194","date":"2025-08-12T05:20:05+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-08-12T04:38:10+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-08-11T17:51:44+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-08-08T06:38:24+00:00","index":"","fulltext":""},{"type":"submitted","content":"npj Unconventional Computing","date":"2025-08-07T12:14:58+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"npj-unconventional-computing","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"Learn more about [npj Unconventional Computing](https://www.nature.com/npjunconvcomput)","snPcode":"44335","submissionUrl":"https://submission.springernature.com/new-submission/44335/3","title":"npj Unconventional Computing","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"NPJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"c486c985-e486-42d9-85f4-164368543c27","owner":[],"postedDate":"August 20th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":53255009,"name":"Physical sciences/Engineering"},{"id":53255010,"name":"Physical sciences/Mathematics and computing"},{"id":53255011,"name":"Biological sciences/Neuroscience"}],"tags":[],"updatedAt":"2025-11-18T21:53:09+00:00","versionOfRecord":[],"versionCreatedAt":"2025-08-20 00:48:43","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7318696","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7318696","identity":"rs-7318696","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}