[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-128186-en":3,"doc-seo-128186-105":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},128186,2336475104042,"Skyler","https://ap-avatar.wpscdn.com/avatar/22000c4c32af1715be0?x-image-process=image/resize,m_fixed,w_180,h_180&k=1786537525561427321",8,"Research & Report","A Quantitative Evaluation of Tiny Machine Learning Models - for Limited-Vocabulary Speech Processing Applications","Tiny Machine Learning (TinyML) enables machine learning on resource-constrained embedded systems, where limited processing power, memory, and energy make deployment of conventional models difficult. This thesis evaluates how common model compression methods—such as quantization, pruning, and weight sharing—affect performance for a limited-vocabulary Arabic speech task focused on the Levantine dialect. Four deep learning architectures (CNN, LSTM, GRU, and BiLSTM) are compared using optimized variants. Experiments deploy models on two different edge devices with distinct constraints to measure memory reduction, accuracy, inference time, and energy efficiency, showing up to 89% memory reduction while preserving accuracy above 97%.","AMERICAN UNIVERSITY OF BEIRUT  \nA Quantitative Evaluation of Tiny Machine Learning Models for Limited-Vocabulary Speech Processing Applications  \nby  \nYASMINE ALI ABU ADLA  \nA thesis  \nsubmitted in partial fulfillment of the requirements for the degree of Master of Engineering  \nto the Department of Electrical and Computer Engineering  \nof Maroun Semaan Faculty of Engineering and Architecture at the American University of Beirut  \nBeirut, Lebanon  \nApril 2023  \nAMERICAN UNIVERSITY OF BEIRUT  \nA Quantitative Evaluation of Tiny Machine Learning Models for Limited-Vocabulary Speech Processing Applications  \nby  \nYASMINE ALI ABU ADLA  \nApproved by:  \n\n| Dr. Mazen Saghir, Associate Professor Electrical and Computer Engineering | Advisor\u003Cbr> |\n| --- | --- |\n| Dr. Mariette Awad, Associate Professor Electrical and Computer Engineering | Co-Advisor |\n| Dr. Jihad Fahs, Assistant Professor Electrical and Computer Engineering | Member of Committee\u003Cbr> |\n\nDr. Wassim El Hajj, Professor Member of Committee  \nComputer Science  \nDate of thesis defense: April 26, 2023  \nAMERICAN UNIVERSITY OF BEIRUT  \nTHESIS RELEASE FORM  \nAbu Adla Yasmine Ali  \nStudent Name:    \nLast First Middle  \nI authorize the American University of Beirut, to: (a) reproduce hard or electronic copies of my thesis; (b) include such copies in the archives and digital repositories of the University; and (c) make freely available such copies to third parties for research or educational purposes  \nX  As of the date of submission of my thesis  \n  After 1 year from the date of submission of my thesis .  \n  After 2 years from the date of submission of my thesis .  \n  After 3 years from the date of submission of my thesis .  \n May 8, 2023  Signature Date  \nAcknowledgements  \nI would like to express my deepest appreciation to my supervisor, Dr. Mazen Saghir and Co-Advisor Dr. Mariette Awad, for their unwavering support, guidance, and encouragement throughout my research journey. Their expertise, insight, and constructive feedback have been invaluable in shaping my ideas and refining my research methodology.  \nI am also grateful to the members of my thesis committee, Dr. Jihad Fahs and Dr. Wassim El Hajj, for their valuable contributions, insightful feedback, and constructive criticism during the development of this thesis. Their collective expertise has played an instrumental role in shaping the direction and scope of my research.  \nFurthermore, I would like to extend my sincere appreciation to my colleagues, friends, and family members who have provided me with unwavering support and encouragement during the challenging times. Their unwavering support, words of wisdom, and kindness have been a constant source of inspiration and motivation throughout my academic journey.  \nFinally, I would like to express my gratitude to the American University of Beirut, which has provided me with the resources, infrastructure, and intellectual environment necessary for the successful completion of my research. I would also like to extend my sincere thanks to the US-Middle East Partnership Initiative (MEPI) for providing me with a full scholarship to participate in The Tomorrow’s Leaders Graduate Program. This program, funded by the U.S. Department of State’s MEPI, not only supported my academic pursuits but also provided me with invaluable leadership development training. Without this scholarship, pursuing graduate studies ina field that I am passionate about would have been nearly impossible. I am grateful for the opportunity to have been a part of this program and for the unwavering support of the MEPI team.  \nThank you all for your invaluable support, encouragement, and guidance throughout this journey.  \nAbstract of the Thesis of  \nYasmine Ali Abu Adla for  Master of Engineering  \nMajor: Electrical and Computer Engineering  \nTitle: A Quantitative Evaluation of Tiny Machine Learning Models for LimitedVocabulary Speech Processing Applications  \nTiny Machine Learning (TinyML) is a rapidly growing fie","cbCaiiT9tfXTJTW1","https://ap.wps.com/l/cbCaiiT9tfXTJTW1","pdf",4578204,5,1,90,"English","en",105,"# Abstract\n## Motivation and TinyML background\n## Compression methods and evaluated models\n## Experimental setup on edge devices\n## Reported results and performance outcomes","[{\"question\":\"What is the goal of this thesis?\",\"answer\":\"The thesis quantifies how compression techniques influence TinyML model performance for a limited-vocabulary speech processing task in Arabic (Levantine dialect).\"},{\"question\":\"Which compression techniques are studied?\",\"answer\":\"It examines methods such as quantization, pruning, and weight sharing to reduce model size and computational cost while maintaining accuracy.\"},{\"question\":\"How are models tested for real-world suitability?\",\"answer\":\"Optimized models are deployed on two resource-constrained edge devices representing different hardware limitations to compare memory footprint, accuracy, inference time, and energy consumption.\"}]","A Quantitative Evaluation of Tiny Machine Learning Models - for Limited-Vocabulary Speech Processing Applications | PDF",1785945377,227,{"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},"a-quantitative-evaluation-of-tiny-machine-learning-models-for-limited-vocabulary-speech-processing-applications","",{"@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/a-quantitative-evaluation-of-tiny-machine-learning-models-for-limited-vocabulary-speech-processing-applications/128186/",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","2026-08-28","2026-08-05",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},"What is the goal of this thesis?","Question",{"text":77,"@type":78},"The thesis quantifies how compression techniques influence TinyML model performance for a limited-vocabulary speech processing task in Arabic (Levantine dialect).","Answer",{"name":80,"@type":75,"acceptedAnswer":81},"Which compression techniques are studied?",{"text":82,"@type":78},"It examines methods such as quantization, pruning, and weight sharing to reduce model size and computational cost while maintaining accuracy.",{"name":84,"@type":75,"acceptedAnswer":85},"How are models tested for real-world suitability?",{"text":86,"@type":78},"Optimized models are deployed on two resource-constrained edge devices representing different hardware limitations to compare memory footprint, accuracy, inference time, and energy consumption.","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},[95,98,102,106,110,115,120,123,128,131,135],{"id":21,"doc_module":4,"doc_module_name":47,"category_name":96,"show_sort_weight":22,"slug":97},"Story & Novel","story-novel",{"id":48,"doc_module":4,"doc_module_name":47,"category_name":99,"show_sort_weight":100,"slug":101},"Literature",80,"literature",{"id":54,"doc_module":4,"doc_module_name":47,"category_name":103,"show_sort_weight":104,"slug":105},"Exam",70,"exam",{"id":20,"doc_module":4,"doc_module_name":47,"category_name":107,"show_sort_weight":108,"slug":109},"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":47,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":47,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":47,"category_name":12,"show_sort_weight":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":47,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":47,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":47,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":47,"category_name":137,"show_sort_weight":20,"slug":138},19,"General","general"]