[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-prototypenas-finds-tiny-neural-nets-for-microcontrollers-in-minutes":10,"sections":35},{"siteName":4,"siteTagline":5,"publisherName":4,"contactEmail":6},"The Revision","Tech news, decoded.","editor@therevision.news",{"gaMeasurementId":8,"adsenseClientId":9},"G-ZW2MV82GYR","ca-pub-8533917693782264",{"article":11},{"id":12,"slug":13,"title":14,"dek":15,"body_md":16,"tags_json":17,"published_at":18,"created_at":19,"updated_at":20,"status":21,"review_note":22,"review_notes":23,"image_url":22,"persona_id":22,"persona_name":22,"section":24,"tags":25,"sources":30,"feedback":34,"feedback_at":22,"cost_usd":34,"total_tokens":34},1782,"prototypenas-finds-tiny-neural-nets-for-microcontrollers-in-minutes","PrototypeNAS Finds Tiny Neural Nets for Microcontrollers in Minutes","A zero-shot architecture search method cuts the time to find deployable neural networks for constrained edge hardware from days to minutes.","Fitting a capable neural network onto a microcontroller just got faster, if not yet easy.\n\nResearchers have released PrototypeNAS, a zero-shot neural architecture search (NAS) method designed to automate the process of finding, compressing, and tuning deep neural networks for microcontroller units (MCUs). Unlike most NAS approaches, which require training dozens or hundreds of candidate models from scratch, PrototypeNAS skips training entirely during the search phase. It works in three steps: carving candidate architectures from a combined search space that spans multiple model types alongside their pruning and quantization settings, using an ensemble of zero-shot proxy scores to rank candidates without training, and then applying Hypervolume subset selection to pick the models that best balance accuracy against compute cost. The researchers tested the approach across 12 datasets covering image classification, time series classification, and object detection.\n\nThe significance here is less about any single result and more about the workflow it replaces. Manual architecture tuning for edge hardware is slow, specialized work - every new MCU target effectively restarts the process. A method that compresses that search to minutes, while accounting for the actual memory and compute limits of the target chip, removes a real bottleneck for teams trying to ship ML on cheap hardware at scale.\n\nThe broader NAS field has been moving toward zero-shot and training-free proxies for a few years now, with mixed results - faster search often trades off against final model quality. PrototypeNAS claims comparable accuracy to larger models, but that comparison will depend heavily on which baselines readers choose to trust.","[\"machine learning\",\"edge ai\",\"neural architecture search\",\"hardware\"]","2026-06-19T04:00:00.000Z","2026-06-19T11:46:32.586Z","2026-06-19T14:22:19.181Z","published",null,[],"ai",[26,27,28,29],"machine learning","edge ai","neural architecture search","hardware",[31],{"name":32,"url":33},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.15106",0,{"sections":36},[37,41,45,50,55,59,64,68,72,77,82,87,92,97],{"name":38,"slug":24,"count":39,"latest_published_at":40},"AI",491,"2026-06-19T14:59:11.000Z",{"name":42,"slug":43,"count":44,"latest_published_at":18},"Security","security",132,{"name":46,"slug":47,"count":48,"latest_published_at":49},"Policy","policy",88,"2026-06-16T09:26:09.000Z",{"name":51,"slug":52,"count":53,"latest_published_at":54},"Consumer Tech","consumer-tech",78,"2026-06-16T17:58:24.000Z",{"name":56,"slug":29,"count":57,"latest_published_at":58},"Hardware",62,"2026-06-18T15:24:16.000Z",{"name":60,"slug":61,"count":62,"latest_published_at":63},"Deals","deals",58,"2026-06-19T14:43:50.000Z",{"name":65,"slug":66,"count":62,"latest_published_at":67},"Software","software","2026-06-16T20:00:00.000Z",{"name":69,"slug":70,"count":71,"latest_published_at":18},"Dev Tools","dev-tools",50,{"name":73,"slug":74,"count":75,"latest_published_at":76},"Science","science",38,"2026-06-18T04:00:00.000Z",{"name":78,"slug":79,"count":80,"latest_published_at":81},"Gaming","gaming",31,"2026-06-16T15:25:13.000Z",{"name":83,"slug":84,"count":85,"latest_published_at":86},"General","general",26,"2026-06-13T18:35:15.000Z",{"name":88,"slug":89,"count":90,"latest_published_at":91},"Startups","startups",23,"2026-06-16T15:00:00.000Z",{"name":93,"slug":94,"count":95,"latest_published_at":96},"Reviews","reviews",19,"2026-06-14T08:00:00.000Z",{"name":98,"slug":99,"count":100,"latest_published_at":101},"How-To","how-to",6,"2026-06-16T09:00:00.000Z"]