[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-128835-105":59,"doc-detail-128835-en":130},{"code":4,"msg":5,"data":6},0,"success",[7,13,18,23,28,33,38,43,48,51,55],{"id":8,"doc_module":4,"doc_module_name":9,"category_name":10,"show_sort_weight":11,"slug":12},1,"Document","Story & Novel",90,"story-novel",{"id":14,"doc_module":4,"doc_module_name":9,"category_name":15,"show_sort_weight":16,"slug":17},2,"Literature",80,"literature",{"id":19,"doc_module":4,"doc_module_name":9,"category_name":20,"show_sort_weight":21,"slug":22},4,"Exam",70,"exam",{"id":24,"doc_module":4,"doc_module_name":9,"category_name":25,"show_sort_weight":26,"slug":27},5,"Comic",60,"comic",{"id":29,"doc_module":4,"doc_module_name":9,"category_name":30,"show_sort_weight":31,"slug":32},6,"Technology",50,"technology",{"id":34,"doc_module":4,"doc_module_name":9,"category_name":35,"show_sort_weight":36,"slug":37},7,"Healthcare",40,"healthcare",{"id":39,"doc_module":4,"doc_module_name":9,"category_name":40,"show_sort_weight":41,"slug":42},8,"Research & Report",30,"research-report",{"id":44,"doc_module":4,"doc_module_name":9,"category_name":45,"show_sort_weight":46,"slug":47},9,"Religion & Spirituality",20,"religion-spirituality",{"id":46,"doc_module":4,"doc_module_name":9,"category_name":49,"show_sort_weight":46,"slug":50},"World Cup","world-cup",{"id":52,"doc_module":4,"doc_module_name":9,"category_name":53,"show_sort_weight":52,"slug":54},10,"Lifestyle","lifestyle",{"id":56,"doc_module":4,"doc_module_name":9,"category_name":57,"show_sort_weight":24,"slug":58},19,"General","general",{"code":4,"msg":60,"data":61},"ok",{"site_id":62,"language":63,"slug":64,"title":65,"keywords":66,"description":67,"schema_data":68,"social_meta":123,"head_meta":125,"extra_data":127,"updated_unix":129},105,"en","efficient-deep-learning-deployments-for-edge-ai-on-resource-constrained-environments-using-hls4ml-diploma-thesis","Efficient Deep Learning deployments for Edge AI on resource-constrained environments using hls4ml - Diploma Thesis","","This diploma thesis investigates Efficient Edge Machine Learning to execute neural network inference near data sources, reducing latency, power consumption, and communication bandwidth. It develops and deploys FPGA and ASIC implementations for two CERN-linked applications: marine pollution detection via on-board image segmentation and pixel-detector signal processing for particle identification and feature extraction. The work extends CERN’s hls4ml library through tool migration, FIFO sizing automation, convolution and MAC optimizations, then reports results on accuracy, resource use, energy, latency, and throughput.",{"@graph":69,"@context":122},[70,84,105],{"@type":71,"itemListElement":72},"BreadcrumbList",[73,77,79,82],{"item":74,"name":75,"@type":76,"position":8},"https://docshare.wps.com","Home","ListItem",{"item":78,"name":9,"@type":76,"position":14},"https://docshare.wps.com/document/",{"item":80,"name":40,"@type":76,"position":81},"https://docshare.wps.com/document/research-report/",3,{"item":83,"name":65,"@type":76,"position":19},"https://docshare.wps.com/document/efficient-deep-learning-deployments-for-edge-ai-on-resource-constrained-environments-using-hls4ml-diploma-thesis/128835/",{"url":83,"name":65,"@type":85,"image":86,"author":91,"headline":65,"publisher":94,"fileFormat":97,"inLanguage":63,"description":67,"dateModified":98,"datePublished":99,"encodingFormat":97,"isAccessibleForFree":100,"interactionStatistic":101},"DigitalDocument",{"url":87,"@type":88,"width":89,"height":90},"https://docshare.wps.com/thumbnails/efficient-deep-learning-deployments-for-edge-ai-on-resource-constrained-environments-using-hls4ml-diploma-thesis/128835.png","ImageObject",300,407,{"name":92,"@type":93},"Maeve","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-09-18","2026-08-06",true,{"@type":102,"interactionType":103,"userInteractionCount":52},"InteractionCounter",{"@type":104},"ViewAction",{"@type":106,"mainEntity":107},"FAQPage",[108,114,118],{"name":109,"@type":110,"acceptedAnswer":111},"What problem does Edge Machine Learning address in this thesis?","Question",{"text":112,"@type":113},"It runs machine learning algorithms on devices near the network edge so data can be processed close to where it is produced, lowering latency, power use, and the amount of data sent to central systems.","Answer",{"name":115,"@type":110,"acceptedAnswer":116},"What are the two main application scenarios explored?",{"text":117,"@type":113},"The thesis covers neural network inference for an Earth observation nano-satellite marine pollution detection use case and a PixESL project integrating neural networks into silicon pixel detectors for particle-related signal analysis.",{"name":119,"@type":110,"acceptedAnswer":120},"Which hls4ml enhancements are emphasized for efficient deployment?",{"text":121,"@type":113},"Key work includes migrating from AMD Vivado HLS to Vitis HLS on a Zynq UltraScale+ platform, automating FIFO queue sizing, optimizing the HLS convolution implementation, and improving MAC operations through DSP FPGA slice packaging.","https://schema.org",{"og:url":83,"og:type":124,"og:title":65,"og:site_name":95,"og:description":67},"article",{"robots":126,"canonical":83},"index,follow",{"doc_id":128,"site_id":62},128835,1786003790,{"code":4,"msg":5,"data":131},{"doc_id":128,"user_id":132,"nickname":92,"user_avatar":133,"doc_module":4,"category_id":39,"category_name":40,"doc_title":65,"doc_description":67,"doc_content":134,"file_id":135,"file_url":136,"file_type":137,"file_size":138,"view_count":52,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":139,"language":140,"language_code":63,"site_id":62,"html_lang":63,"table_of_contents":141,"faqs":142,"seo_title":143,"seo_description":67,"update_tm":129,"read_time":144},2336474466712,"https://ap-avatar.wpscdn.com/davatar_a8503ba1806abce46bf441b54a3ca4cd","National Technical University of Athens School of Electrical and Computer Engineering Division of Computer Science  \nEfficient Deep Learning deployments for Edge AI on resource-constrained environments  \nusing hls4ml  \nDiploma Thesis  \nof  \nSTYLIANOS TZELEPIS  \nSupervisor: Dionisios Pnevmatikatos  \nProfessor, ECE NTUA  \nAthens, April 2025  \nNational Technical University of Athens School of Electrical and Computer Engineering Division of Computer Science  \nEfficient Deep Learning deployments for Edge AI on resource-constrained environments using hls4ml  \nDiploma Thesis  \nof  \nSTYLIANOS TZELEPIS  \nSupervisor: Dionisios Pnevmatikatos  \nProfessor, ECE NTUA  \nApproved by the examination committee on 9th April 2025 .  \nDionisios Pnevmatikatos Nectarios Koziris Theodoros Alexopoulos  \nProfessor, ECE NTUA Professor, ECE NTUA Professor, AMPS NTUA  \nAthens, April 2025  \nNational Technical University of Athens School of Electrical and Computer Engineering Division of Computer Science  \nCopyright © – All rights reserved.  \nStylianos Tzelepis, 2025 .  \nThe copying, storage and distribution of this diploma thesis, exall or part ofit, is prohibited for commercial purposes. Reprinting, storage and distribution for non-profit, educational or of a research nature is allowed, provided that the source is indicated and that this message is retained.  \nThe content of this thesis does not necessarily reflect the views of the Department, the Supervisor, or the committee that approved it.  \nDISCLAIMER ON ACADEMIC ETHICS AND INTELLECTUAL PROPERTY RIGHTS  \nBeing fully aware of the implications of copyright laws, I expressly state that this diploma thesis, as well as the electronic files and source codes developed or modified in the course of this thesis, are solely the product of my personal work and do not infringe any rights of intellectual property, personality and personal data of third parties, do not contain work / contributions of third parties for which the permission of the authors / beneficiaries is required and are not a product of partial or complete plagiarism, while the sources used are limited to the bibliographic references only and meet the rules of scientific citing. The points where I have used ideas, text, files and / or sources of other authors are clearly mentioned in the text with the appropriate citation and the relevant complete reference is included in the bibliographic references section. I fully, individually and personally undertake all legal and administrative consequences that may arise in the event that it is proven, in the course of time, that this thesis or part of it does not belong to me because it is a product of plagiarism.  \nStylianos Tzelepis 26th February 2025  \nAbstract  \nThe present thesis was carried out within the framework of the CERN Technical Student program under the guidance of Professor Dionysios Pnevmatikatos, Professor of the Faculty of Electrical and Computer Engineering at the National Technical University of Athens and Drs. Sioni Paris Summers and Maurizio Pierini, CERN researchers in the CMS experiment at CERN.  \nEdge Machine Learning (EML) involves running Machine Learning algorithms on devices located near the edges of a data network. This technique allows data to be processed close to the source, thus reducing network latency, power consumption, and the amount of data that needs to be transferred to central systems.  \nThis paper focuses on two EML applications, involving the implementation and execution of Neural Networks on FPGA (Field-Programmable Gate Arrays) and ASIC (ApplicationSpecific Integrated Circuit) devices, respectively.  \nIn the context of the Edge SpAIce project, the objective is to run Convolutional Neural Networks on an Earth observation nano-satellite for marine pollution detection. In conventional Earth Observation missions, all the payload data are sent to Earth without prior processing. However, the pollution phenomenon, like other similar phenomena, occurs rarely compared ","cbCaih3st8R58Q5G","https://ap.wps.com/l/cbCaih3st8R58Q5G","pdf",12066752,91,"English","# Abstract\n## Edge Machine Learning overview\n## Application 1: Earth observation pollution detection\n## hls4ml-based deployment on FPGA/ASIC\n## Application improvements and optimization workflow\n## Application 2: PixESL pixel-detector neural integration\n## Evaluated outcomes and performance metrics","[{\"question\":\"What problem does Edge Machine Learning address in this thesis?\",\"answer\":\"It runs machine learning algorithms on devices near the network edge so data can be processed close to where it is produced, lowering latency, power use, and the amount of data sent to central systems.\"},{\"question\":\"What are the two main application scenarios explored?\",\"answer\":\"The thesis covers neural network inference for an Earth observation nano-satellite marine pollution detection use case and a PixESL project integrating neural networks into silicon pixel detectors for particle-related signal analysis.\"},{\"question\":\"Which hls4ml enhancements are emphasized for efficient deployment?\",\"answer\":\"Key work includes migrating from AMD Vivado HLS to Vitis HLS on a Zynq UltraScale+ platform, automating FIFO queue sizing, optimizing the HLS convolution implementation, and improving MAC operations through DSP FPGA slice packaging.\"}]","Efficient Deep Learning deployments for Edge AI on resource-constrained environments using hls4ml - Diploma Thesis | PDF",229]