[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-121655-en":3,"doc-seo-121655-105":30,"detail-sidebar-cat-0-en-105":90},{"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":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":27,"seo_description":14,"update_tm":28,"read_time":29},121655,1099513958762,"Logic","https://ap-avatar.wpscdn.com/avatar/1000023916a998db790?x-image-process=image/resize,m_fixed,w_180,h_180&k=1784791008015729253",6,"Technology","Towards Battery-Free Machine Learning Inference and Model Personalization on MCUs - Poster","Machine learning is increasingly moving to edge devices, but resource-constrained environments still limit conventional inference due to high compute, memory, and energy demands. The work presents a battery-free ML inference and model personalization pipeline for microcontroller units (MCUs) and demonstrates it with fish image recognition in the ocean. Experiments compare optimized versus baseline models using accuracy, runtime, power, and energy metrics. Results reach 97.78% accuracy with 483.82 KB Flash, 70.32 KB RAM, 118 ms runtime, 4.83 mW power, and 0.57 mJ energy, significantly reducing resource usage.","Poster: Towards Battery-Free Machine Learning Inference and  \nModel Personalization on MCUs  \nYushan Huang Imperial College London  \nLondon, UK [yushan.huang21@imperial.ac.uk](yushan.huang21@imperial.ac.uk)  \nHamed Haddadi Imperial College London  \nLondon, UK [h.haddadi@imperial.ac.uk](h.haddadi@imperial.ac.uk)  \narXiv :2305 . 18954v1 [ cs .LG] 30 May 2023  \nABSTRACT  \nMachine learning (ML) is moving towards edge devices. However, ML models with high computational demands and energy consumption pose challenges for ML inference in resource-constrained environments, such as the deep sea. To address these challenges, we propose a battery-free ML inference and model personalization pipeline for microcontroller units (MCUs) . As an example, we performed fish image recognition in the ocean. We evaluated and compared the accuracy, runtime, power, and energy consumption of the model before and after optimization. The results demonstrate that, our pipeline can achieve 97.78% accuracy with 483.82 KB Flash, 70.32 KB RAM, 118 ms runtime, 4.83 mW power, and 0.57 mJ energy consumption on MCUs, reducing by 64.17%, 12.31%, 52.42%, 63.74%, and 82.67%, compared to the baseline. The results indicate the feasibility of battery-free ML inference on MCUs.  \nCCS CONCEPTS  \n• Computing methodologies → Computer vision; • Computer systems organization → Embedded systems.  \nKEYWORDS  \nEdge Computing, IoT, TinyML, Resource-constrained  \nACM Reference Format:  \nYushan Huang and Hamed Haddadi. 2023. Poster: Towards Battery-Free Machine Learning Inference and Model Personalization on MCUs. In The 21st Annual International Conference on Mobile Systems, Applications and Services (MobiSys ’23), June 18–22, 2023, Helsinki, Finland. ACM, New York, NY, USA, 2 pages. [https://doi.org/10.1145/3581791.3597371](https://doi.org/10.1145/3581791.3597371)  \n1 INTRODUCTION  \nMachine learning (ML) models have become ubiquitous in solving diverse problems. However, these models often demand high memory, computation power, and energy requirements, posing challenges for ML deployment on resource-constrained edge devices. The edge offers several advantages such as reduced response latency, better bandwidth utilization, and improved security and privacy expectations. Therefore, there is a pressing need to develop optimized lightweight ML models for deployment on the edge.  \nPermission to make digital or hard copies of part or all ofthis work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for third-party components of this work must be honored. For all other uses, contact the owner/author(s) .  \nMobiSys ’23, June 18–22, 2023, Helsinki, Finland © 2023 Copyright held by the owner/author(s) .  \nACM ISBN 979-8-4007-0110-8/23/06 .  \n[https://doi.org/10.1145/3581791.3597371](https://doi.org/10.1145/3581791.3597371)  \nFigure 1: Design of battery-free inference on MCUs  \nTo address this problem, researchers have explored model compression techniques to compress the model for the edge [1] . However, these technologies typically assume that the edge has sufficient memory, computing power, and power supply, which can be challenging to achieve in extreme environments such as deep seas, and remote areas. Some studies have deployed traditional models such as SVM on MCUs [2], but they require manual feature extraction and may not perform well on high-dimensional data. Additionally, power and energy consumption are often overlooked in edge-based ML deployment. Recent advancements in battery-free sensing technology have allowed for innovative applications [3] . These techniques have made long-term MCUs use possible without the need for power supplies or batteries. this study aims to examine the feasibility of achieving battery-free ML inference and model personalization on MCUs for extreme environments.  \nIn our previous wo","cbCainF4dWmyxOQF","https://ap.wps.com/l/cbCainF4dWmyxOQF","pdf",1432023,1,2,"English","en",105,"# Abstract\n# Introduction\n# Pipeline\n## Preprocessing\n## Modeling\n## Quantization and Deployment and Inference","[{\"question\":\"What problem does the poster address for ML on resource-constrained MCUs?\",\"answer\":\"It targets the difficulty of running energy- and compute-intensive ML inference on devices with limited memory, computation, and power supply, especially in extreme environments like the deep sea.\"},{\"question\":\"How does the proposed method enable battery-free ML inference?\",\"answer\":\"It provides an end-to-end battery-free ML inference and model personalization pipeline for MCUs, including preprocessing, modeling, quantization, and deployment/inference.\"},{\"question\":\"What performance improvements are reported after optimization on MCUs?\",\"answer\":\"The optimized pipeline reports 97.78% accuracy with 483.82 KB Flash, 70.32 KB RAM, 118 ms runtime, 4.83 mW power, and 0.57 mJ energy, with substantial reductions versus the baseline.\"}]","Towards Battery-Free Machine Learning Inference and Model Personalization on MCUs - Poster | PDF",1785805978,5,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":85,"head_meta":87,"extra_data":89,"updated_unix":28},"towards-battery-free-machine-learning-inference-and-model-personalization-on-mcus-poster","",{"@graph":36,"@context":84},[37,53,67],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,47,50],{"item":41,"name":42,"@type":43,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":21},"https://docshare.wps.com/document/","Document",{"item":48,"name":12,"@type":43,"position":49},"https://docshare.wps.com/document/technology/",3,{"item":51,"name":13,"@type":43,"position":52},"https://docshare.wps.com/document/towards-battery-free-machine-learning-inference-and-model-personalization-on-mcus-poster/121655/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":23,"description":14,"dateModified":61,"datePublished":61,"encodingFormat":60,"isAccessibleForFree":62,"interactionStatistic":63},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":41,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-08-04",true,{"@type":64,"interactionType":65,"userInteractionCount":20},"InteractionCounter",{"@type":66},"ViewAction",{"@type":68,"mainEntity":69},"FAQPage",[70,76,80],{"name":71,"@type":72,"acceptedAnswer":73},"What problem does the poster address for ML on resource-constrained MCUs?","Question",{"text":74,"@type":75},"It targets the difficulty of running energy- and compute-intensive ML inference on devices with limited memory, computation, and power supply, especially in extreme environments like the deep sea.","Answer",{"name":77,"@type":72,"acceptedAnswer":78},"How does the proposed method enable battery-free ML inference?",{"text":79,"@type":75},"It provides an end-to-end battery-free ML inference and model personalization pipeline for MCUs, including preprocessing, modeling, quantization, and deployment/inference.",{"name":81,"@type":72,"acceptedAnswer":82},"What performance improvements are reported after optimization on MCUs?",{"text":83,"@type":75},"The optimized pipeline reports 97.78% accuracy with 483.82 KB Flash, 70.32 KB RAM, 118 ms runtime, 4.83 mW power, and 0.57 mJ energy, with substantial reductions versus the baseline.","https://schema.org",{"og:url":51,"og:type":86,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":88,"canonical":51},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":91},[92,96,100,104,108,111,116,121,126,129,133],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":93,"show_sort_weight":94,"slug":95},"Story & Novel",90,"story-novel",{"id":21,"doc_module":4,"doc_module_name":46,"category_name":97,"show_sort_weight":98,"slug":99},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":46,"category_name":101,"show_sort_weight":102,"slug":103},"Exam",70,"exam",{"id":29,"doc_module":4,"doc_module_name":46,"category_name":105,"show_sort_weight":106,"slug":107},"Comic",60,"comic",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":109,"slug":110},50,"technology",{"id":112,"doc_module":4,"doc_module_name":46,"category_name":113,"show_sort_weight":114,"slug":115},7,"Healthcare",40,"healthcare",{"id":117,"doc_module":4,"doc_module_name":46,"category_name":118,"show_sort_weight":119,"slug":120},8,"Research & Report",30,"research-report",{"id":122,"doc_module":4,"doc_module_name":46,"category_name":123,"show_sort_weight":124,"slug":125},9,"Religion & Spirituality",20,"religion-spirituality",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":127,"show_sort_weight":124,"slug":128},"World Cup","world-cup",{"id":130,"doc_module":4,"doc_module_name":46,"category_name":131,"show_sort_weight":130,"slug":132},10,"Lifestyle","lifestyle",{"id":134,"doc_module":4,"doc_module_name":46,"category_name":135,"show_sort_weight":29,"slug":136},19,"General","general"]