[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-117326-en":3,"doc-seo-117326-105":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},117326,1099514068035,"Ezra","https://ap-avatar.wpscdn.com/davatar_276721f389ce27ea32af1340a28f341c",8,"Research & Report","Machine Learning in Container Orchestration Systems: Applications and Deployment - Doctor of Philosophy Thesis","Machine learning methods, especially reinforcement learning, are examined for resource management tasks including resource allocation and scheduling optimization in cloud environments. The thesis targets Kubernetes-driven Mobile Edge Clouds (MEC), proposing Mobile-Kube and Smart-Kube to learn migration and scheduling policies that improve trade-offs between latency minimization and energy conservation beyond traditional heuristic methods. It further develops Kubernetes-based, cost-efficient machine learning systems for edge-cloud continuum deployments via InfAdapter and IPA autoscaling approaches that incorporate inference accuracy alongside resource costs, demonstrating improved balance of accuracy and cost versus state-of-the-art approaches.","Machine Learning in Container Orchestration Systems: Applications and Deployment  \nSaeid Ghafouri  \nSupervisor: Dr. Joseph Doyle Submitted in partial fulfilment of the requirements of the Degree of Doctor of Philosophy  \nQueen Mary University of London School of Electronic Engineering and Computer Science  \nUnited Kingdom  \nNovember 2023  \nStatement of Originality  \nI, Saeid Ghafouri, confirm that the research included within this thesis is my own work or that where it has been carried out in collaboration with, or supported by others, that this is duly acknowledged below and my contribution indicated. Previously published material is also acknowledged below.  \nI attest that I have exercised reasonable care to ensure that the work is original, and does not to the best of my knowledge break any UK law, infringe any third party’s copyright or other Intellectual Property Right, or contain any confidential material.  \nI accept that Queen Mary University of London has the right to use plagiarism detection software to check the electronic version of the thesis. I confirm that this thesis has not been previously submitted for the award of a degree by this or any other university.  \nThe copyright of this thesis rests with the author and no quotation from it or information derived from it may be published without the prior written consent of the author.  \nSignature: Saeid Ghafouri  \nDate: 4/1/2024  \nDetails of collaboration and publications:  \n• Chapter 5 of this thesis was a joint work between the first author and the author of this thesis, while the main credit goes to the first author, the author of this thesis has helped with the idea, supervision, and implementation which sums up to a total of 30% contribution in the authorship.  \nAbstract  \nIn recent years, machine learning methods, particularly Reinforcement Learning, have become increasingly popular for addressing resource management challenges, notably in resource allocation and optimizing scheduling processes. Kubernetes, a widely adopted container orchestration platform, plays a crucial role in enabling effective resource management within the realm of cloud computing. This thesis focuses on specific scheduling scenarios within Kubernetes-driven Mobile Edge Clouds (MEC) and seeks to identify appropriate machine learning methods to address resource management challenges across diverse cloud computing domains, including edge and fog computing, harnessing the strengths of Reinforcement Learning. The two proposed solutions, Mobile-Kube and Smart-Kube, employ Reinforcement Learning to acquire distinct migration and scheduling policies. These policies aim to strike improved balances between competing objectives such as minimizing latency and conserving energy. Results indicate that both systems successfully learn policies capable of managing these trade-offs more effectively than traditional heuristic approaches.  \nApplications in the edge cloud continuum that perform machine learning operations are usually supported through the machine learning models deployed on specialized infrastructure optimized for machine learning services. Another focus of this thesis is developing cost-efficient and accurate machine learning systems in Kubernetes. Specific features of machine learning applications like prediction accuracy necessitate the development of specialized infrastructures for the deployment of such services. These infrastructures must provide scalable computing resources and containerization support. The two proposed systems, InfAdapter and IPA, introduce innovative autoscaling methods utilizing various machine learning model variants. These methods not only factor in resource costs but also take into account the accuracy of machine learning inference services. The findings suggest that this new optimization approach can effectively manage multiple trade-offs between accuracy and cost objectives, outperforming current state-of-the-art methods.  \nIn today’s cloud-native landscape, ","cbCaidK0bjz1Dm2X","https://ap.wps.com/l/cbCaidK0bjz1Dm2X","pdf",5545042,1,193,"English","en",105,"# Abstract\n## Resource management with reinforcement learning in Kubernetes MEC\n## Cost-efficient machine learning deployment and autoscaling in Kubernetes\n## Cloud-native Kubernetes and containerization for ML systems","[{\"question\":\"What resource management problems does the thesis address with machine learning?\",\"answer\":\"It focuses on resource allocation and scheduling optimization challenges, using reinforcement learning to derive effective migration and scheduling policies.\"},{\"question\":\"What are Mobile-Kube and Smart-Kube designed to do?\",\"answer\":\"They use reinforcement learning to learn distinct migration and scheduling policies, aiming for better trade-offs between latency reduction and energy conservation in Kubernetes-driven MEC.\"},{\"question\":\"How do InfAdapter and IPA improve Kubernetes-based ML deployments?\",\"answer\":\"They introduce autoscaling methods that consider both resource costs and the inference accuracy of machine learning services, improving the accuracy-cost trade-off compared with existing approaches.\"}]","Machine Learning in Container Orchestration Systems: Applications and Deployment - Doctor of Philosophy Thesis | PDF",1785675205,486,{"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},"machine-learning-in-container-orchestration-systems-applications-and-deployment-doctor-of-philosophy-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/machine-learning-in-container-orchestration-systems-applications-and-deployment-doctor-of-philosophy-thesis/117326/",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 resource management problems does the thesis address with machine learning?","Question",{"text":75,"@type":76},"It focuses on resource allocation and scheduling optimization challenges, using reinforcement learning to derive effective migration and scheduling policies.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"What are Mobile-Kube and Smart-Kube designed to do?",{"text":80,"@type":76},"They use reinforcement learning to learn distinct migration and scheduling policies, aiming for better trade-offs between latency reduction and energy conservation in Kubernetes-driven MEC.",{"name":82,"@type":73,"acceptedAnswer":83},"How do InfAdapter and IPA improve Kubernetes-based ML deployments?",{"text":84,"@type":76},"They introduce autoscaling methods that consider both resource costs and the inference accuracy of machine learning services, improving the accuracy-cost trade-off compared with existing approaches.","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"]