[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-81852-en":3,"doc-seo-81852-105":31,"detail-sidebar-cat-0-en-105":93},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":21,"is_downloadable":21,"audit_status":21,"page_count":22,"language":23,"language_code":24,"site_id":25,"html_lang":24,"table_of_contents":26,"faqs":27,"seo_title":28,"seo_description":14,"update_tm":29,"read_time":30},81852,5909877438554,"Maeve","https://ap-avatar.wpscdn.com/avatar/5600025385ad2bf12a7?_k=1778553567797529272",8,"Research & Report","Towards transferable lightweight neuromorphic computing through a model-free temporal-switch framework","Lightweight neuromorphic computing enables efficient AI for resource-constrained edge deployments, yet scalable transfer is blocked by unavoidable device-to-device variations that force costly retraining and calibration. A model-free temporal-switch (TS) framework is proposed to improve direct readout transfer without post-training adjustment. TS incorporates broader device diversity during training, validated with memristor-based reservoir computing on unseen devices, improved Mackey–Glass prediction, 92.4% accuracy for spoken digit classification, and consistent results across memristor families and RC configurations.","arXiv :2607 .02608v 1 [ cs .AR] 1 Jul 2026  \nTowards transferable lightweight neuromorphic computing through a model-free temporal-switch  \nframework  \nZefeng Zhang 1,2,3 , Chao Li 1,3 , Siyao Chen2,4 , Pei Chen 1,3 , Bo-Wei Qin2,5,6*, Xumeng Zhang 1,3*, Wei Lin2,4,5,6*, Qi Liu 1,3*  \n1 State Key Laboratory of Integrated Chips and Systems, Fudan University, Shanghai, 200433, China.  \n2 Research Institute of Intelligent Complex Systems, Fudan University, Shanghai, 200433, China.  \n3 Frontier Institute of Chip and System, Fudan University, Shanghai, 200433, China.  \n4 School of Mathematical Sciences and Shanghai Center for Mathematical Sciences, Shanghai, 200433, China.  \n5 Shanghai Artificial Intelligence Laboratory, Shanghai, 200232, China.  \n6 State Key Laboratory of Medical Neurobiology and MOE Frontiers Center for Brain Science, Institute of Brain Science, Fudan University,  \nShanghai, 200032, China.  \n*Corresponding author(s). E-mail(s): [boweiqin@fudan.edu.cn](boweiqin@fudan.edu.cn) ; [xumengzhang@fudan.edu.cn](xumengzhang@fudan.edu.cn) ; [wlin@fudan.edu.cn](wlin@fudan.edu.cn) ; qi [liu@fudan.edu.cn](liu@fudan.edu.cn) ;  \nContributing authors: [21110850044@m.fudan.edu.cn](21110850044@m.fudan.edu.cn) ; [lichao@ime.ac.cn](lichao@ime.ac.cn) ;  \n[21210180115@m.fudan.edu.cn](21210180115@m.fudan.edu.cn) ; [22112020077@m.fudan.edu.cn](22112020077@m.fudan.edu.cn) ;  \nAbstract  \nLightweight neuromorphic computing offers a promising route to efficient AI, with particular benefits for resource-constrained edge deployments. However, its scalable deployment that can reliably transfer the expected performance has long been hindered by device-to-device variations, which necessitate costly and repeated re-training on new copies and undermine the practical advantages. To address this issue, we introduce a model-free temporal-switch (TS) framework to improve the direct transfer performance, without post-training calibration or adjustment. The TS framework provides a methodology to incorporate a broader  \n1  \nspectrum of devices in the training process. In the validation using memristorbased reservoir computing, it enables high performance on unseen devices with a directly transferred readout. It achieves improved prediction in the representative Mackey–Glass benchmark, and the accuracy of 92.4% in spoken digit classification. Its efficacy is validated across different memristor families and RC configurations. Theoretical analysis not only reveals the general computational mechanism underlying its efficacy, but also underlines its potential applicability to other physical platforms.  \nKeywords: neuromorphic computing, memristor, reservoir computing, device variation  \nIntroduction  \nLightweight neuromorphic computing has attracted increasing interest in both academia and industry as a promising hardware-oriented route to efficient AI [1–3], particularly in resource-constrained edge scenarios where many distributed devices operate in parallel [4, 5] . Among diverse lightweight neuromorphic computing paradigms, physical reservoir computing (RC) is particularly attractive [6, 7] . It uses the intrinsic nonlinear and short-term memory dynamics of physical devices [8–12], such as dynamic memristors [13, 14], as computational black-boxes (‘brokenisomorphism’ [15]) . By employing a simple three-layer architecture where only the readout is trained [16], it enables useful computation based on measured states [17, 18] while reducing the need for accurate device modelling and extensive parameter tuning [19, 20] . Consequently, it offers an efficient and scalable way to harness the rich device dynamics for information processing.  \nHowever, for such systems to move beyond individually optimized demonstrations towards reliably deployable hardware, the trained readout should be transferable across nominally identical hardware instances rather than tied to one specific physical device [21, 22] . It is a central and persistent challenge because of the wide","cbCaimTLhxo2C6Nk","https://ap.wps.com/l/cbCaimTLhxo2C6Nk","pdf",3068333,9,1,28,"English","en",105,"# Abstract\n# Introduction\n## Lightweight neuromorphic and physical reservoir computing\n## Transferability challenge from device-to-device variation\n## Temporal-switch framework for improved direct transfer","[{\"question\":\"为什么轻量级神经形态计算的性能难以在不同器件间可靠迁移？\",\"answer\":\"由于器件到器件（D2D）变化，输入到响应状态的映射、状态协方差以及状态-目标相关性都会改变，导致在某个物理储备库上训练的读出对其他名义相同但未见过的器件直接迁移时性能下降。\"},{\"question\":\"该文提出的模型无关时间开关（TS）框架如何提升读出迁移能力？\",\"answer\":\"TS框架通过在训练过程中引入更广泛的器件谱系信息，从而让读出更具跨器件可迁移性，并实现无需后训练校准或调整的直接迁移效果。\"},{\"question\":\"文中如何验证TS框架的有效性？\",\"answer\":\"在基于忆阻器的储备计算（reservoir computing）验证中，TS框架在未见器件上可直接迁移读出实现高性能；同时在Mackey–Glass基准上提高预测表现，并在口语数字分类中达到92.4%准确率，且跨不同忆阻器家族与RC配置均保持有效。\"}]","Towards transferable lightweight neuromorphic computing through a model-free temporal-switch framework | PDF",1784176652,71,{"code":4,"msg":32,"data":33},"ok",{"site_id":25,"language":24,"slug":34,"title":13,"keywords":35,"description":14,"schema_data":36,"social_meta":88,"head_meta":90,"extra_data":92,"updated_unix":29},"towards-transferable-lightweight-neuromorphic-computing-through-a-model-free-temporal-switch-framework","",{"@graph":37,"@context":87},[38,55,70],{"@type":39,"itemListElement":40},"BreadcrumbList",[41,45,49,52],{"item":42,"name":43,"@type":44,"position":21},"https://docshare.wps.com","Home","ListItem",{"item":46,"name":47,"@type":44,"position":48},"https://docshare.wps.com/document/","Document",2,{"item":50,"name":12,"@type":44,"position":51},"https://docshare.wps.com/document/research-report/",3,{"item":53,"name":13,"@type":44,"position":54},"https://docshare.wps.com/document/towards-transferable-lightweight-neuromorphic-computing-through-a-model-free-temporal-switch-framework/81852/",4,{"url":53,"name":13,"@type":56,"author":57,"headline":13,"publisher":59,"fileFormat":62,"inLanguage":24,"description":14,"dateModified":63,"datePublished":64,"encodingFormat":62,"isAccessibleForFree":65,"interactionStatistic":66},"DigitalDocument",{"name":9,"@type":58},"Person",{"url":42,"name":60,"@type":61},"DocShare","Organization","application/pdf","2026-08-03","2026-07-16",true,{"@type":67,"interactionType":68,"userInteractionCount":20},"InteractionCounter",{"@type":69},"ViewAction",{"@type":71,"mainEntity":72},"FAQPage",[73,79,83],{"name":74,"@type":75,"acceptedAnswer":76},"为什么轻量级神经形态计算的性能难以在不同器件间可靠迁移？","Question",{"text":77,"@type":78},"由于器件到器件（D2D）变化，输入到响应状态的映射、状态协方差以及状态-目标相关性都会改变，导致在某个物理储备库上训练的读出对其他名义相同但未见过的器件直接迁移时性能下降。","Answer",{"name":80,"@type":75,"acceptedAnswer":81},"该文提出的模型无关时间开关（TS）框架如何提升读出迁移能力？",{"text":82,"@type":78},"TS框架通过在训练过程中引入更广泛的器件谱系信息，从而让读出更具跨器件可迁移性，并实现无需后训练校准或调整的直接迁移效果。",{"name":84,"@type":75,"acceptedAnswer":85},"文中如何验证TS框架的有效性？",{"text":86,"@type":78},"在基于忆阻器的储备计算（reservoir computing）验证中，TS框架在未见器件上可直接迁移读出实现高性能；同时在Mackey–Glass基准上提高预测表现，并在口语数字分类中达到92.4%准确率，且跨不同忆阻器家族与RC配置均保持有效。","https://schema.org",{"og:url":53,"og:type":89,"og:title":13,"og:site_name":60,"og:description":14},"article",{"robots":91,"canonical":53},"index,follow",{"doc_id":7,"site_id":25},{"code":4,"msg":5,"data":94},[95,99,103,107,112,117,122,125,129,132,136],{"id":21,"doc_module":4,"doc_module_name":47,"category_name":96,"show_sort_weight":97,"slug":98},"Story & Novel",90,"story-novel",{"id":48,"doc_module":4,"doc_module_name":47,"category_name":100,"show_sort_weight":101,"slug":102},"Literature",80,"literature",{"id":54,"doc_module":4,"doc_module_name":47,"category_name":104,"show_sort_weight":105,"slug":106},"Exam",70,"exam",{"id":108,"doc_module":4,"doc_module_name":47,"category_name":109,"show_sort_weight":110,"slug":111},5,"Comic",60,"comic",{"id":113,"doc_module":4,"doc_module_name":47,"category_name":114,"show_sort_weight":115,"slug":116},6,"Technology",50,"technology",{"id":118,"doc_module":4,"doc_module_name":47,"category_name":119,"show_sort_weight":120,"slug":121},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":47,"category_name":12,"show_sort_weight":123,"slug":124},30,"research-report",{"id":20,"doc_module":4,"doc_module_name":47,"category_name":126,"show_sort_weight":127,"slug":128},"Religion & Spirituality",20,"religion-spirituality",{"id":127,"doc_module":4,"doc_module_name":47,"category_name":130,"show_sort_weight":127,"slug":131},"World Cup","world-cup",{"id":133,"doc_module":4,"doc_module_name":47,"category_name":134,"show_sort_weight":133,"slug":135},10,"Lifestyle","lifestyle",{"id":137,"doc_module":4,"doc_module_name":47,"category_name":138,"show_sort_weight":108,"slug":139},19,"General","general"]