[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-117327-en":3,"doc-seo-117327-105":30,"detail-sidebar-cat-0-en-105":91},{"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":4,"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},117327,1099514068035,"Ezra","https://ap-avatar.wpscdn.com/davatar_276721f389ce27ea32af1340a28f341c",8,"Research & Report","Supporting Distributed Machine Learning in Heterogeneous and Dynamic Environment - Master of Science Thesis","Growing demand for Internet-of-Things applications drives interest in executing machine learning workflows across heterogeneous edge devices through distributed systems. Meeting higher computational performance while maintaining accuracy is difficult because distributed environments introduce complexity and reliability constraints. This thesis develops an adaptive algorithm for distributed machine learning in heterogeneous, dynamic settings. Devices are configured with different epoch counts based on prior-round training time. Results indicate improved execution time without accuracy loss compared with federated averaging.","SUPPORTING DISTRIBUTED MACHINE LEARNING IN HETEROGENEOUS AND  \nDYNAMIC ENVIRONMENT  \nBy  \nMOHD SHAFIQAZIZAN  \nA thesis submitted in partial fulfillment of the requirements for the degree of  \nMASTER OF SCIENCE IN COMPUTER SCIENCE  \nWASHINGTON STATE UNIVERSITY  \nSchool of Engineering and Computer Science, Vancouver  \nDECEMBER 2023  \n©Copyright by MOHD SHAFIQAZIZAN, 2023 All Rights Reserved  \n©Copyright by MOHD SHAFIQAZIZAN, 2023 All Rights Reserved  \nTo the Faculty of Washington State University:  \nThe members of the Committee appointed to examine the thesis of MOHD SHAFIQAZIZAN find it satisfactory and recommend that it be accepted.  \nXinghui Zhao, Ph.D., Chair Xuechen Zhang, Ph.D. Scott Wallace, Ph.D.  \nACKNOWLEDGEMENTS  \nI want to acknowledge and warmly thank my advisor, Dr. Xinghui Zhao, who made this work possible. Her guidance and continuous advice carried me through all the stages of writing my thesis. I extend my most incredible gratitude to my committee members, Dr. Xuechen Zhang and Dr. Scott Wallace, for making my defense enjoyable and for the comments and suggestions; thanks to both.  \nNext, I would like to thank my employer, the Malaysian Agricultural Research and Development Institute (MARDI), for sponsoring my master’s degree program and providing the trust and opportunity to send me abroad to complete my study. I sincerely thank my parents for their continuous support and understanding when undertaking my research and writing my thesis. Your prayer for me was what sustained me this far.  \nLastly, I would like to thank God for letting me through all the difficulties. I have experienced your guidance day by day. You are the one who let me finish my degree, and I will keep on trusting you for my future.  \nSUPPORTING DISTRIBUTED MACHINE LEARNING IN HETEROGENEOUS AND  \nDYNAMIC ENVIRONMENT  \nAbstract  \nby Mohd ShafiqAzizan, M.S.  \nWashington State University  \nDecember 2023  \nChair: Xinghui Zhao  \nWith the growing popularity demand of Internet-of-Things applications, there is high interest in supporting machine learning workflows using heterogeneous edge devices, i.e., in a distributed system. Nonetheless, achieving better computational performance without sacrificing the accuracy in distributed machine learning is challenging due to the complexity and reliability of distributed systems. In this study, we support distributed machine learning in heterogeneous and dynamic environments.  \nTo overcome these impediments, we have successfully developed an adaptive algorithm that enhances computational performance in distributed machine learning. The edge devices were dynamically configured with different number of epoch depending on their execution training time from the previous round. Our analysis showed promising performance of the adaptive algorithm for better execution time without sacrificing the accuracy compared to the federated averaging algorithm.  \nTABLE OF CONTENTS  \nPage  \nACKNOWLEDGEMENTS .................................. iii  \nABSTRACT ........................................... iv  \nLIST OF FIGURES ....................................... vii  \nLIST OF TABLES ........................................ viii  \nCHAPTER  \n1. INTRODUCTION ................................... 1  \nProblem Statement and Research Questions ................... 2  \nPurpose of the Study ................................. 5  \n2. RELATED WORK ................................... 6  \nDistributed ML Framework ............................. 6  \nHeterogeneous and Dynamic Environment in Distributed ML ........ 7  \n3. PRELIMINARY STUDY ................................ 9  \nFlower Framework .................................. 9  \nHyperparameter impacts on training execution time .............. 12  \nLearning Rate ................................... 12  \nEpoch ........................................ 13  \nBatch Size ..................................... 13  \nRounds ....................................... 14  \n4. ADAPTIVE TUNING ALGORITHM .................","cbCaimotqerboymZ","https://ap.wps.com/l/cbCaimotqerboymZ","pdf",3072589,1,64,"English","en",105,"# Abstract\n# Acknowledgements\n# List of Figures\n# List of Tables\n# 1. Introduction\n## Problem Statement and Research Questions\n## Purpose of the Study\n# 2. Related Work\n## Distributed ML Framework\n## Heterogeneous and Dynamic Environment in Distributed ML\n# 3. Preliminary Study\n## Flower Framework\n## Hyperparameter impacts on training execution time\n# 4. Adaptive Tuning Algorithm\n## Adaptive Tuning Algorithm\n## Custom Strategy Implementation\n## System Implementation\n# 5. Result and Discussion\n## System Evaluation Design\n## Experimental Setup Design\n## Experimental Result\n# 6. Conclusion and Future Work\n# Bibliography\n# Appendix A. List of Result","[{\"question\":\"What challenge does this thesis address in distributed machine learning?\",\"answer\":\"Achieving better computational performance in distributed machine learning is challenging without sacrificing accuracy, due to the complexity and reliability constraints of distributed systems.\"},{\"question\":\"How does the proposed adaptive algorithm improve training performance?\",\"answer\":\"Edge devices are dynamically configured with different epoch counts based on their execution training time from the previous round, aiming to enhance execution time while preserving accuracy.\"},{\"question\":\"What comparison is used to evaluate the adaptive algorithm?\",\"answer\":\"Performance is compared against the federated averaging algorithm to assess execution time improvements without accuracy loss.\"}]","Supporting Distributed Machine Learning in Heterogeneous and Dynamic Environment - Master of Science Thesis | PDF",1785675213,161,{"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":86,"head_meta":88,"extra_data":90,"updated_unix":28},"supporting-distributed-machine-learning-in-heterogeneous-and-dynamic-environment-master-of-science-thesis","",{"@graph":36,"@context":85},[37,54,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/supporting-distributed-machine-learning-in-heterogeneous-and-dynamic-environment-master-of-science-thesis/117327/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":62,"encodingFormat":61,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-02",true,{"@type":65,"interactionType":66,"userInteractionCount":4},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"What challenge does this thesis address in distributed machine learning?","Question",{"text":75,"@type":76},"Achieving better computational performance in distributed machine learning is challenging without sacrificing accuracy, due to the complexity and reliability constraints of distributed systems.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How does the proposed adaptive algorithm improve training performance?",{"text":80,"@type":76},"Edge devices are dynamically configured with different epoch counts based on their execution training time from the previous round, aiming to enhance execution time while preserving accuracy.",{"name":82,"@type":73,"acceptedAnswer":83},"What comparison is used to evaluate the adaptive algorithm?",{"text":84,"@type":76},"Performance is compared against the federated averaging algorithm to assess execution time improvements without accuracy loss.","https://schema.org",{"og:url":52,"og:type":87,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":89,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,123,128,131,135],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":46,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":46,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":46,"category_name":137,"show_sort_weight":106,"slug":138},19,"General","general"]