Dynamic Token Masking in Spiking Neural Network

Dynamic Token Masking in Spiking Neural Network | 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 Dynamic Token Masking in Spiking Neural Network Yuetong Fang, Ziqing Wang, Deming Zhou, Hongwei Ren, Shibo Zhou, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6004117/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Spiking Neural Networks (SNNs) offer an energy-efficient alternative to conventional artificial neural networks (ANNs). Although ANN-to-SNN conversion has emerged as a promising direction for developing high-performance spike-driven models with reduced training complexity, it still faces the challenge of maintaining energy efficiency in converted SNNs. In this paper, we address these limitations by introducing a dynamic spiking token mixer, inspired by the strong information redundancy present in the spike self-attention mechanism. Our approach effectively replicates the selective processing capabilities of self-attention through dynamic token masking (DynMask), with layer-specific masking ratios customized according to both spatial and temporal significance. Comprehensive results establish DynMask as a practical step toward efficient deep learning systems. Specifically, DynMask achieves performance gains of up to +3.23% on ImageNet-1K while narrowing the accuracy gaps to as low as +0.02% compared to ANNs, with simultaneous energy consumption reductions of up to 44×. Moreover, our approach successfully extends to complex vision tasks that remain largely unexplored in SNN literature, including COCO detection and ADE20K segmentation. Computational Neuroscience Computer Architecture and Engineering Neuromorphic Computing Spiking Neural Network Token Mixing Full Text Additional Declarations The authors declare no competing interests. Cite Share Download PDF Status: Posted 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-6004117","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":414017825,"identity":"1366dd90-fd9b-42ba-81ac-52c7f0b7ae62","order_by":0,"name":"Yuetong Fang","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Yuetong","middleName":"","lastName":"Fang","suffix":""},{"id":414017826,"identity":"b3dda56b-c1b5-4264-b734-c2b050e466a5","order_by":1,"name":"Ziqing Wang","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Ziqing","middleName":"","lastName":"Wang","suffix":""},{"id":414017827,"identity":"0f306007-fb93-450c-a2d6-864dcbdef0b9","order_by":2,"name":"Deming Zhou","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Deming","middleName":"","lastName":"Zhou","suffix":""},{"id":414017828,"identity":"7918a013-8fb7-4e85-80dd-8c37d2dd471a","order_by":3,"name":"Hongwei Ren","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Hongwei","middleName":"","lastName":"Ren","suffix":""},{"id":414017829,"identity":"d15b5be6-4551-4c42-a047-9a3d42eb502c","order_by":4,"name":"Shibo Zhou","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Shibo","middleName":"","lastName":"Zhou","suffix":""},{"id":414017830,"identity":"bb283ae5-bf29-42ba-a754-b764ccfd1a10","order_by":5,"name":"Renjing Xu","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABDElEQVRIie2PsU7DMBCGL4qULM6eChS/wlmW6MI78AqJskaIiQmKUSWzFHgBeIeywWZkiS5WZ6ouZU8lVkSGOmXI5CgjUv0NJ511391vAI/nn4J5W+NZ2D2RQQoxrYLDlD/SaqAyvrtXF1+v18V8tdWcNA3QMxVsthLo2KEcm2WOhVlwXJ+XZSIRmFIhe5bA3oQjTlohFvIjw3XFdSLw5uVWREeJhByVQ6H1XiG4MlyTxl6ZQvzbq6SkVa4y/CS8JBECjSAKexWyD6b4yFTcfoEDkmA6elqmbO5SYnPCfuSkeFgYntZNBvRRv3/Xl6fUdcViw4DuWlSBaHc55y3hBmDStVT0DXs8Hs8hsgPfl1UUEHT/RAAAAABJRU5ErkJggg==","orcid":"","institution":"","correspondingAuthor":true,"prefix":"","firstName":"Renjing","middleName":"","lastName":"Xu","suffix":""}],"badges":[],"createdAt":"2025-02-11 06:08:58","currentVersionCode":1,"declarations":{"humanSubjects":false,"vertebrateSubjects":false,"conflictsOfInterestStatement":false,"humanSubjectEthicalGuidelines":false,"humanSubjectConsent":false,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":false},"doi":"10.21203/rs.3.rs-6004117/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6004117/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":76089107,"identity":"e2849033-ea32-43e6-acd0-dcea4f666eeb","added_by":"auto","created_at":"2025-02-12 08:11:48","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":3591186,"visible":true,"origin":"","legend":"","description":"","filename":"IJCVsubmission.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6004117/v1_covered_a82cc841-cbc8-490b-b603-1831c4a8a23b.pdf"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"\u003cp\u003eDynamic Token Masking in Spiking Neural Network\u003c/p\u003e","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"Hong Kong University of Science and Technology","isAcceptedByJournal":false,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Neuromorphic Computing, Spiking Neural Network, Token Mixing","lastPublishedDoi":"10.21203/rs.3.rs-6004117/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6004117/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eSpiking Neural Networks (SNNs) offer an energy-efficient alternative to conventional artificial neural networks (ANNs). Although ANN-to-SNN conversion has emerged as a promising direction for developing high-performance spike-driven models with reduced training complexity, it still faces the challenge of maintaining energy efficiency in converted SNNs. In this paper, we address these limitations by introducing a dynamic spiking token mixer, inspired by the strong information redundancy present in the spike self-attention mechanism. Our approach effectively replicates the selective processing capabilities of self-attention through dynamic token masking (DynMask), with layer-specific masking ratios customized according to both spatial and temporal significance. Comprehensive results establish DynMask as a practical step toward efficient deep learning systems. Specifically, DynMask achieves performance gains of up to +3.23% on ImageNet-1K while narrowing the accuracy gaps to as low as +0.02% compared to ANNs, with simultaneous energy consumption reductions of up to 44×. Moreover, our approach successfully extends to complex vision tasks that remain largely unexplored in SNN literature, including COCO detection and ADE20K segmentation.\u003c/p\u003e","manuscriptTitle":"Dynamic Token Masking in Spiking Neural Network","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-02-12 08:03:36","doi":"10.21203/rs.3.rs-6004117/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"eef18445-902f-4b2b-bf6c-51f4353416fd","owner":[],"postedDate":"February 12th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":44146046,"name":"Computational Neuroscience"},{"id":44146047,"name":"Computer Architecture and Engineering"}],"tags":[],"updatedAt":"2025-02-12T08:03:36+00:00","versionOfRecord":[],"versionCreatedAt":"2025-02-12 08:03:36","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6004117","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6004117","identity":"rs-6004117","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}