[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-seo-290350-105":3,"detail-sidebar-cat-0-en-105":81,"doc-detail-290350-en":130},{"code":4,"msg":5,"data":6},0,"ok",{"site_id":7,"language":8,"slug":9,"title":10,"keywords":11,"description":12,"schema_data":13,"social_meta":74,"head_meta":76,"extra_data":78,"updated_unix":80},105,"en","middleware-systems-for-efficient-and-robust-distributed-machine-learning-dissertation","Middleware Systems for Efficient and Robust Distributed Machine Learning - Dissertation","","A dissertation presenting middleware systems that improve efficiency and robustness for distributed machine learning across training, continual adaptation, and inference. It frames real-world deployment challenges such as data heterogeneity, platform heterogeneity and stragglers, and distribution shift. The work develops middleware approaches for participant selection in federated learning, introduces signal-driven mechanisms to handle joint heterogeneity, and proposes shift-aware mixture-of-experts adaptation under streaming changes, emphasizing resilience and overhead tradeoffs.",{"@graph":14,"@context":73},[15,34,56],{"@type":16,"itemListElement":17},"BreadcrumbList",[18,23,27,31],{"item":19,"name":20,"@type":21,"position":22},"https://docshare.wps.com","Home","ListItem",1,{"item":24,"name":25,"@type":21,"position":26},"https://docshare.wps.com/document/","Document",2,{"item":28,"name":29,"@type":21,"position":30},"https://docshare.wps.com/document/research-report/","Research & Report",3,{"item":32,"name":10,"@type":21,"position":33},"https://docshare.wps.com/document/middleware-systems-for-efficient-and-robust-distributed-machine-learning-dissertation/290350/",4,{"url":32,"name":10,"@type":35,"image":36,"author":41,"headline":10,"publisher":44,"fileFormat":47,"inLanguage":8,"description":12,"dateModified":48,"datePublished":49,"encodingFormat":47,"isAccessibleForFree":50,"interactionStatistic":51},"DigitalDocument",{"url":37,"@type":38,"width":39,"height":40},"https://docshare.wps.com/thumbnails/middleware-systems-for-efficient-and-robust-distributed-machine-learning-dissertation/290350.png","ImageObject",300,407,{"name":42,"@type":43},"Eliana","Person",{"url":19,"name":45,"@type":46},"DocShare","Organization","application/pdf","2026-10-19","2026-09-17",true,{"@type":52,"interactionType":53,"userInteractionCount":55},"InteractionCounter",{"@type":54},"ViewAction",9,{"@type":57,"mainEntity":58},"FAQPage",[59,65,69],{"name":60,"@type":61,"acceptedAnswer":62},"What problem does the dissertation address in distributed machine learning deployments?","Question",{"text":63,"@type":64},"It addresses challenges in real-world deployments, including data heterogeneity, platform heterogeneity with stragglers, and distribution shift that degrade training and adaptation quality.","Answer",{"name":66,"@type":61,"acceptedAnswer":67},"How does the proposed middleware improve federated learning robustness during training?",{"text":68,"@type":64},"It uses signal-driven middleware for participant selection under joint heterogeneity, leveraging label distribution as a signal and applying straggler-aware selection and overprovisioning.",{"name":70,"@type":61,"acceptedAnswer":71},"What approach is used for continual adaptation when distributions shift over time?",{"text":72,"@type":64},"It introduces ShiftEx, a shift-aware mixture-of-experts framework that assigns experts at the party level and supports adaptation under streaming distribution changes.","https://schema.org",{"og:url":32,"og:type":75,"og:title":10,"og:site_name":45,"og:description":12},"article",{"robots":77,"canonical":32},"index,follow",{"doc_id":79,"site_id":7},290350,1789641443,{"code":4,"msg":82,"data":83},"success",[84,88,92,96,101,106,111,115,119,122,126],{"id":22,"doc_module":4,"doc_module_name":25,"category_name":85,"show_sort_weight":86,"slug":87},"Story & Novel",90,"story-novel",{"id":26,"doc_module":4,"doc_module_name":25,"category_name":89,"show_sort_weight":90,"slug":91},"Literature",80,"literature",{"id":33,"doc_module":4,"doc_module_name":25,"category_name":93,"show_sort_weight":94,"slug":95},"Exam",70,"exam",{"id":97,"doc_module":4,"doc_module_name":25,"category_name":98,"show_sort_weight":99,"slug":100},5,"Comic",60,"comic",{"id":102,"doc_module":4,"doc_module_name":25,"category_name":103,"show_sort_weight":104,"slug":105},6,"Technology",50,"technology",{"id":107,"doc_module":4,"doc_module_name":25,"category_name":108,"show_sort_weight":109,"slug":110},7,"Healthcare",40,"healthcare",{"id":112,"doc_module":4,"doc_module_name":25,"category_name":29,"show_sort_weight":113,"slug":114},8,30,"research-report",{"id":55,"doc_module":4,"doc_module_name":25,"category_name":116,"show_sort_weight":117,"slug":118},"Religion & Spirituality",20,"religion-spirituality",{"id":117,"doc_module":4,"doc_module_name":25,"category_name":120,"show_sort_weight":117,"slug":121},"World Cup","world-cup",{"id":123,"doc_module":4,"doc_module_name":25,"category_name":124,"show_sort_weight":123,"slug":125},10,"Lifestyle","lifestyle",{"id":127,"doc_module":4,"doc_module_name":25,"category_name":128,"show_sort_weight":97,"slug":129},19,"General","general",{"code":4,"msg":82,"data":131},{"doc_id":79,"user_id":132,"nickname":42,"user_avatar":133,"doc_module":4,"category_id":112,"category_name":29,"doc_title":10,"doc_description":12,"doc_content":134,"file_id":135,"file_url":136,"file_type":137,"file_size":138,"view_count":55,"is_deleted":4,"is_public":22,"is_downloadable":22,"audit_status":22,"page_count":139,"language":140,"language_code":8,"site_id":7,"html_lang":8,"table_of_contents":141,"faqs":142,"seo_title":143,"seo_description":12,"update_tm":80,"read_time":144,"preview_page_count":4},4398048949847,"https://ap-avatar.wpscdn.com/avatar/400002536579ef2da7f?_k=1778318612642679267","UC Irvine  \nUC Irvine Electronic Theses and Dissertations  \nTitle  \nMiddleware Systems for Efficient and Robust Distributed Machine Learning  \nPermalink  \n[https://escholarship.org/uc/item/3m35k8qt](https://escholarship.org/uc/item/3m35k8qt)  \nISBN  \n9798244893700  \nAuthor  \nBhope, Rahul Atul  \nPublication Date  \n2026-05-21  \nCopyright Information  \nThis work is made available under the terms of a Creative Commons Attribution License, available at [https://creativecommons.org/licenses/by/4.0/](https://creativecommons.org/licenses/by/4.0/)  \n[Peer reviewed|Thesis/dissertation](Peer reviewed|Thesis/dissertation)  \n[eScholarship.org](eScholarship.org) Powered by the California Digital Library  \nUniversity of California  \nUNIVERSITY OF CALIFORNIA,  \nIRVINE  \nMiddleware Systems for Efficient and Robust Distributed Machine Learning  \nDISSERTATION  \nsubmitted in partial satisfaction of the requirements  \nfor the degree of  \nDOCTOR OF PHILOSOPHY  \nin Networked Systems  \nby  \nRahul Atul Bhope  \nDissertation Committee:  \nProfessor Nalini Venkatasubramanian, UCI, Chair Professor Marco Levorato, UCI  \nProfessor Sangeetha Abdu Jyothi, UCI Professor Sharad Mehrotra, UCI  \nDr. Jayaram K. Radhakrishnan, IBM Research  \nChapter 4 © 2023 ACM/IFIP International Middleware Conference  \nChapter 5 © 2025 ACM/IFIP International Middleware Conference Portion of Chapter 6 © 2025 Conference on Empirical Methods in Natural Language  \nProcessing (EMNLP) All other materials © 2026 Rahul Atul Bhope  \nTABLE OF CONTENTS  \nPage  \nLIST OF FIGURES vi  \nLIST OF TABLES ix  \nLIST OF ALGORITHMS xi  \nACKNOWLEDGMENTS xii  \nVITA xiii  \nABSTRACT OF THE DISSERTATION xv  \n1 Introduction 1  \n1.1 The Machine Learning Lifecycle and the Role of Middleware ......... 2  \n1.2 Challenges in Real-World Deployments ..................... 4  \n1.3 Problem Statement ................................ 6  \n1.4 Thesis Contributions ............................... 7  \n1.5 Dissertation Organization ............................ 8  \n2 Related Work 10  \n2.1 Systems for Distributed and Adaptive ML ................... 11  \n2.1.1 Distributed Training Systems and the Parameter-Server Line ..... 11  \n2.1.2 Scheduling, Placement, and Resource Allocation ............ 12  \n2.1.3 Edge Learning, Federated Learning, and Decentralized Paradigms .. 14  \n2.2 Continual Learning and Adaptation under Distribution Shift ......... 15  \n2.2.1 Catastrophic Forgetting and Continual Learning ............ 16  \n2.2.2 Adaptation under Distribution Shift: Domain Adaptation and TestTime Adaptation ............................. 17  \n2.2.3 Distributed, Edge, and Federated Learning under Shift and Drift ... 19  \n2.3 Inference-Time Optimization in Large Language and Vision-Language Models 20  \n2.3.1 Decoding, reranking, and single-model adaptive computation ..... 21  \n2.3.2 Routing, Cascades, MoE-Style Sparsity, and Test-Time Scaling Tradeoffs ..................................... 23  \n2.3.3 Retrieval pruning, long-context selection, and memory-aware inference 24  \n2.3.4 Benchmarks vs. Real-World System Constraints ............ 26  \n3 The Approach: Middleware for the ML Lifecycle 28  \n3.1 Chapter Overview ................................. 28  \n3.2 Motivation ..................................... 29  \n3.3 The Middleware Layer .............................. 30  \n3.4 Middleware Across the ML Lifecycle ...................... 32  \n3.4.1 Training: Participant Selection under Data Heterogeneity ....... 32  \n3.4.2 Continual Adaptation: Responding to Distribution Shift ....... 32  \n3.4.3 Inference: Decision Making under Resource Constraints ........ 33  \n3.5 Summary ..................................... 34  \n4 Robust Training under Data and Platform Heterogeneity 35  \n4.1 Chapter Overview ................................. 35  \n4.2 Background .................................... 36  \n4.2.1 Data Heterogeneity in Federated Learning ............... 36  \n4.2.2 Platform Heterogeneity and Stragglers ................. 38  \n4.2.3 The Compoundin","cbCaij9iUnJu0jc4","https://ap.wps.com/l/cbCaij9iUnJu0jc4","pdf",4646010,167,"English","# List of Figures\n# List of Tables\n# List of Algorithms\n# Acknowledgments\n# Vita\n# Abstract of the Dissertation\n# 1 Introduction\n## 1.1 The Machine Learning Lifecycle and the Role of Middleware\n## 1.2 Challenges in Real-World Deployments\n## 1.3 Problem Statement\n## 1.4 Thesis Contributions\n## 1.5 Dissertation Organization\n# 2 Related Work\n## 2.1 Systems for Distributed and Adaptive ML\n## 2.2 Continual Learning and Adaptation under Distribution Shift\n## 2.3 Inference-Time Optimization in Large Language and Vision-Language Models\n# 3 The Approach: Middleware for the ML Lifecycle\n## 3.1 Chapter Overview\n## 3.2 Motivation\n## 3.3 The Middleware Layer\n## 3.4 Middleware Across the ML Lifecycle\n## 3.5 Summary\n# 4 Robust Training under Data and Platform Heterogeneity\n## 4.1 Chapter Overview\n## 4.2 Background\n## 4.3 Related Work\n## 4.4 FLIPS: Signal-Driven Middleware for Federated Learning\n## 4.5 Experimental Evaluation\n## 4.6 Summary and Transition\n# 5 Continual Adaptation under Distribution Shifts\n## 5.1 Chapter Overview\n## 5.2 Background and Challenges\n## 5.3 Related Work\n## 5.4 ShiftEx: Shift-Aware Mixture of Experts Framework","[{\"question\":\"What problem does the dissertation address in distributed machine learning deployments?\",\"answer\":\"It addresses challenges in real-world deployments, including data heterogeneity, platform heterogeneity with stragglers, and distribution shift that degrade training and adaptation quality.\"},{\"question\":\"How does the proposed middleware improve federated learning robustness during training?\",\"answer\":\"It uses signal-driven middleware for participant selection under joint heterogeneity, leveraging label distribution as a signal and applying straggler-aware selection and overprovisioning.\"},{\"question\":\"What approach is used for continual adaptation when distributions shift over time?\",\"answer\":\"It introduces ShiftEx, a shift-aware mixture-of-experts framework that assigns experts at the party level and supports adaptation under streaming distribution changes.\"}]","Middleware Systems for Efficient and Robust Distributed Machine Learning - Dissertation | PDF",421]