[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-128568-en":3,"doc-seo-128568-105":31,"detail-sidebar-cat-0-en-105":92},{"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},128568,549768064622,"Anda","https://ap-avatar.wpscdn.com/davatar_6f874abed73319feea01a86fa6f0fab8",8,"Research & Report","Exploring Probabilistic Models for Semi-supervised Learning","Deep neural networks increasingly power computer vision, but training requires large labeled datasets that are costly to obtain. Semi-supervised learning (SSL) reduces labeling effort by leveraging both labeled and unlabeled data. While leading SSL methods are often deterministic, this thesis investigates probabilistic alternatives to provide uncertainty estimates for safer deployment. It addresses pseudo-label errors and incorrect predictions, and uses uncertainty for filtering unreliable pseudo-labels. Building on MC dropout, the work proposes GBDL, NP-Match, and NP-SemiSeg, delivering improved segmentation, classification, uncertainty quantification, and efficiency.","Exploring Probabilistic Models for Semi-supervised Learning  \nJianfeng Wang  \nLinacre College University of Oxford  \nA thesis submitted for the degree of Doctor of Philosophy  \nTrinity 2023  \nDeclarations  \nI solemnly affirm that, unless explicitly acknowledged, the contents of this dissertation are original and have not been submitted, in whole or in part, for the pursuit of anyother degree or qualification at this University or elsewhere. This dissertation is my own work and contains nothing which is the outcome of work done in collaboration with others, barring those explicitly specified in the text.  \nJianfeng Wang May 2023  \nAcknowledgements  \nFirstly, I extend my profound gratitude to my supervisor, Prof. Thomas Lukasiewicz. His steadfast support, ceaseless encouragement, and inexhaustible patience have been my pillar throughout my DPhil studies. With each progression of my project, he has prioritized my research interests, for which I am immensely grateful. I cherish his invaluable support, which has been present from the inception of my DPhil project when I was still discerning the direction of my research.  \nSecondly, I wish to express my heartfelt appreciation to my family. Their unwavering encouragement, especially amidst the hurdles of daily life and the challenges of my research, has been my solace. I consider myself immensely fortunate to have such supportive parents who have stood by every decision I’ve made. To my maternal grandparents, I owe a deep debt of gratitude for their financial and moral support that has fortified my resilience, allowing me to overcome any obstacle. To my paternal grandparents, who watch over me from heaven, I hope that my accomplishments have made you proud.  \nThirdly, I wish to acknowledge my research collaborators: Xiaolin Hu, Daniela Massiceti, Vladimir Pavlovic, Jianfei Cai, and Alexandros Neophytou. Their generous provision of computational resources and invaluable feedback on my research have been instrumental in my progress.  \nI am also deeply indebted to my peers at the Intelligent Systems Lab for creating a warm and enthusiastic environment.  \nLastly, I wish to thank my friends at Oxford for the endless moments of joy. A special mention goes to my good friend, Yuedong Chen, whose regular, casual conversations have been my stress reliever.  \nAbstract  \nDeep neural networks are increasingly harnessed for computer vision tasks, thanks to their robust performance. However, their training demands large-scale labeled datasets, which are labor-intensive to prepare. Semi-supervised learning (SSL) offers a solution by learning from a mix of labeled and unlabeled data.  \nWhile most state-of-the-art SSL methods follow a deterministic approach, the exploration of their probabilistic counterparts remains limited. This research area is important because probabilistic models can provide uncertainty estimates critical for real-world applications. For instance, SSL-trained models may fall short of those trained with supervised learning due to potential pseudo-label errors in unlabeled data, and these models are more likely to make wrong predictions in practice. Especially in critical sectors like medical image analysis and autonomous driving, decision-makers must understand the model’s limitations and when incorrect predictions may occur, insights often provided by uncertainty estimates. Furthermore, uncertainty can also serve as a criterion for filtering out unreliable pseudo-labels when unlabeled samples are used for training, potentially improving deep model performance.  \nThis thesis furthers the exploration of probabilistic models for SSL. Drawing on the widely-used Bayesian approximation tool, Monte Carlo (MC) dropout, I propose a new probabilistic framework, the Generative Bayesian Deep Learning (GBDL) architecture, for semi-supervised medical image segmentation. This approach not only mitigates potential overfitting found in previous methods but also achieves superior results across f","cbCaigmIVWAW3nnr","https://ap.wps.com/l/cbCaigmIVWAW3nnr","pdf",5917111,3,1,127,"English","en",105,"# Abstract\n## Motivation and problem setting\n## Proposed probabilistic frameworks\n### GBDL for semi-supervised medical image segmentation\n### NP-Match for semi-supervised image classification\n### NP-SemiSeg for semi-supervised semantic segmentation\n# Publications","[{\"question\":\"Why are probabilistic models important for semi-supervised learning in this thesis?\",\"answer\":\"Probabilistic models supply uncertainty estimates that help understand model limitations and identify when incorrect predictions may occur. They also support filtering unreliable pseudo-labels to improve performance.\"},{\"question\":\"How does the thesis build on MC dropout?\",\"answer\":\"It uses the widely used Bayesian approximation tool, Monte Carlo (MC) dropout, as a basis to develop new probabilistic architectures for SSL tasks.\"},{\"question\":\"What are the main contributions proposed for different SSL tasks?\",\"answer\":\"The thesis introduces GBDL for semi-supervised medical image segmentation, NP-Match for large-scale semi-supervised image classification, and NP-SemiSeg for semi-supervised semantic segmentation, each aimed at improving accuracy, uncertainty quantification, and speed.\"}]","Exploring Probabilistic Models for Semi-supervised Learning | PDF",1786001788,320,{"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":87,"head_meta":89,"extra_data":91,"updated_unix":29},"exploring-probabilistic-models-for-semi-supervised-learning","",{"@graph":37,"@context":86},[38,54,69],{"@type":39,"itemListElement":40},"BreadcrumbList",[41,45,49,51],{"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":20},"https://docshare.wps.com/document/research-report/",{"item":52,"name":13,"@type":44,"position":53},"https://docshare.wps.com/document/exploring-probabilistic-models-for-semi-supervised-learning/128568/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":24,"description":14,"dateModified":62,"datePublished":63,"encodingFormat":61,"isAccessibleForFree":64,"interactionStatistic":65},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":42,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-24","2026-08-06",true,{"@type":66,"interactionType":67,"userInteractionCount":20},"InteractionCounter",{"@type":68},"ViewAction",{"@type":70,"mainEntity":71},"FAQPage",[72,78,82],{"name":73,"@type":74,"acceptedAnswer":75},"Why are probabilistic models important for semi-supervised learning in this thesis?","Question",{"text":76,"@type":77},"Probabilistic models supply uncertainty estimates that help understand model limitations and identify when incorrect predictions may occur. They also support filtering unreliable pseudo-labels to improve performance.","Answer",{"name":79,"@type":74,"acceptedAnswer":80},"How does the thesis build on MC dropout?",{"text":81,"@type":77},"It uses the widely used Bayesian approximation tool, Monte Carlo (MC) dropout, as a basis to develop new probabilistic architectures for SSL tasks.",{"name":83,"@type":74,"acceptedAnswer":84},"What are the main contributions proposed for different SSL tasks?",{"text":85,"@type":77},"The thesis introduces GBDL for semi-supervised medical image segmentation, NP-Match for large-scale semi-supervised image classification, and NP-SemiSeg for semi-supervised semantic segmentation, each aimed at improving accuracy, uncertainty quantification, and speed.","https://schema.org",{"og:url":52,"og:type":88,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":90,"canonical":52},"index,follow",{"doc_id":7,"site_id":25},{"code":4,"msg":5,"data":93},[94,98,102,106,111,116,121,124,129,132,136],{"id":21,"doc_module":4,"doc_module_name":47,"category_name":95,"show_sort_weight":96,"slug":97},"Story & Novel",90,"story-novel",{"id":48,"doc_module":4,"doc_module_name":47,"category_name":99,"show_sort_weight":100,"slug":101},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":47,"category_name":103,"show_sort_weight":104,"slug":105},"Exam",70,"exam",{"id":107,"doc_module":4,"doc_module_name":47,"category_name":108,"show_sort_weight":109,"slug":110},5,"Comic",60,"comic",{"id":112,"doc_module":4,"doc_module_name":47,"category_name":113,"show_sort_weight":114,"slug":115},6,"Technology",50,"technology",{"id":117,"doc_module":4,"doc_module_name":47,"category_name":118,"show_sort_weight":119,"slug":120},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":47,"category_name":12,"show_sort_weight":122,"slug":123},30,"research-report",{"id":125,"doc_module":4,"doc_module_name":47,"category_name":126,"show_sort_weight":127,"slug":128},9,"Religion & Spirituality",20,"religion-spirituality",{"id":127,"doc_module":4,"doc_module_name":47,"category_name":130,"show_sort_weight":127,"slug":131},"World Cup","world-cup",{"id":133,"doc_module":4,"doc_module_name":47,"category_name":134,"show_sort_weight":133,"slug":135},10,"Lifestyle","lifestyle",{"id":137,"doc_module":4,"doc_module_name":47,"category_name":138,"show_sort_weight":107,"slug":139},19,"General","general"]