[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-118089-en":3,"doc-seo-118089-105":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},118089,137441390410,"Hazel","https://ap-avatar.wpscdn.com/avatar/2000252f4ab5702993?_k=1776741390130283984",8,"Research & Report","Machine Learning for Histopathology Images in Low-Data Regimes - Doctor of Philosophy Dissertation","The dissertation presents machine-learning methods for histopathology image understanding under low-data regimes. It develops approaches for caption generation, including metric learning via triplet loss with data augmentation and cohort processing, and evaluates performance on real datasets with captioning, clustering, and visualization analyses. It further proposes a hyperbolic attention model based on Poincaré ball geometry to improve representation learning on histopathology image benchmarks. Finally, it introduces set-level label learning to enhance instance-level classification from coarse-grained labels, and presents deep Bayesian active learning for efficient acquisition in limited-label settings.","THE UNIVERSITY OF CHICAGO  \nMACHINE LEARNING FOR HISTOPATHOLOGY IMAGES IN LOW-DATA REGIMES  \nA DISSERTATION SUBMITTED TO  \nTHE FACULTY OF THE DIVISION OF THE PHYSICAL SCIENCES  \nIN CANDIDACY FOR THE DEGREE OF  \nDOCTOR OF PHILOSOPHY  \nDEPARTMENT OF COMPUTER SCIENCE  \nBY  \nRENYU ZHANG  \nCHICAGO, ILLINOIS  \nAUGUST 2024  \nCopyright © 2024 by Renyu Zhang All Rights Reserved  \nTo my beloved wife, Na Chen, whose steadfast confidence in my abilities and unwavering support are the cornerstones of every achievement, continually motivating me to improve. To my parents, Hengxiang Xu and Qingjie Zhang, whose unconditional love and tireless dedication have shaped my values and inspire me to strive for greatness.  \nTABLE OF CONTENTS  \nLIST OF FIGURES .................................... viii  \nLIST OF TABLES ..................................... xi  \nACKNOWLEDGMENTS ................................. xii  \nABSTRACT ........................................ xv  \n1 INTRODUCTION ................................... 1  \n1.1 Motivation ..................................... 3  \n1.2 Summary of Contributions ............................ 4  \n1.3 Thesis Organization ................................ 8  \n1.4 List of Publications ................................ 9  \n2 BACKGROUND AND RELATED WORKS ..................... 11  \n2.1 Histopathology Images .............................. 11  \n2.2 Caption Prediction ................................ 12  \n2.3 Active Learning .................................. 13  \n2.4 Few-shot Learning ................................ 15  \n2.5 Self-supervised Learning ............................. 17  \n3 CAPTION GENERATION FOR HISTOPATHOLOGY IMAGES ......... 20  \n3.1 Motivation ..................................... 20  \n3.2 Methods ...................................... 22  \n3.2.1 Overview ................................. 22  \n3.2.2 Metric Learning with Triplet Loss .................... 23  \n3.2.3 Neural Network Architecture ....................... 25  \n3.2.4 Data Augmentation and Hyperparameter Settings ........... 27  \n3.2.5 Cohort ................................... 27  \n3.3 Results on Real Data ............................... 29  \n3.3.1 Results on Caption Generation ..................... 29  \n3.3.2 Results on Metric Learning ....................... 29  \n3.3.3 Results on Clustering ........................... 30  \n3.3.4 Results on Visualization ......................... 31  \n3.4 Discussion ..................................... 32  \n4 HYPERBOLIC ATTENTION MODEL FOR HISTOPATHOLOGY IMAGES .. 35  \n4.1 Motivation ..................................... 35  \n4.2 Method ...................................... 37  \n4.2.1 Poincaré Ball Model ........................... 38  \n4.2.2 Möbius Addition ............................. 38  \n4.2.3 Exponential and Logarithmic Maps ................... 38  \n4.2.4 Hyperbolic Linear Layer ......................... 39  \n4.2.5 Klein Model ................................ 39  \n4.2.6 Hyperbolic Attention ........................... 39  \n4.2.7 Multiclass Logistic Regression ...................... 40  \n4.3 Results ....................................... 40  \n4.3.1 Camelyon16 ................................ 41  \n4.3.2 TCGA ................................... 43  \n4.4 Discussion and conclusion ............................ 44  \n5 ENHANCING INSTANCE-LEVEL IMAGE CLASSIFICATION WITH SET-LEVEL LABELS ......................................... 45  \n5.1 Motivation ..................................... 45  \n5.2 Fine-Grained Representation Learning from Coarse-Grained Labels ..... 48  \n5.2.1 The FACILE Algorithm ......................... 49  \n5.2.2 Theoretical Analysis ........................... 51  \n5.3 Results ....................................... 53  \n5.3.1 Baseline Models and Algorithm Instantiation .............. 53  \n5.3.2 Pretrain with Unique Class Number of Input Sets ........... 54  \n5.3.3 Pretrain with Most Frequent Class Label ................ 55  \n5.3.4 Fine-tune CLIP Model with Anomaly Detection Dataset .......","cbCaikSEkichSgen","https://ap.wps.com/l/cbCaikSEkichSgen","pdf",48080091,1,181,"English","en",105,"# Introduction\n## Motivation\n## Summary of Contributions\n## Thesis Organization\n## List of Publications\n# Background and Related Works\n## Histopathology Images\n## Caption Prediction\n## Active Learning\n## Few-shot Learning\n## Self-supervised Learning\n# Caption Generation for Histopathology Images\n## Motivation\n## Methods\n## Results on Real Data\n## Discussion\n# Hyperbolic Attention Model for Histopathology Images\n## Motivation\n## Method\n## Results\n## Discussion and conclusion\n# Enhancing Instance-level Image Classification with Set-level Labels\n## Motivation\n## Fine-Grained Representation Learning from Coarse-Grained Labels\n## Results\n## Related Work\n## Conclusion and Discussion\n# Deep Bayesian Active Learning\n## Motivation\n## Problem Setup\n## Our Approach\n## Experiments\n## Conclusion\n# Conclusion and Outlook","[{\"question\":\"What problem does the dissertation address about histopathology images?\",\"answer\":\"It focuses on machine learning for histopathology image understanding when labeled data is limited, targeting multiple tasks including caption generation and instance-level classification with set-level labels.\"},{\"question\":\"Which model is introduced to leverage non-Euclidean geometry?\",\"answer\":\"The dissertation proposes a hyperbolic attention model for histopathology images, built using Poincaré ball geometry concepts such as Möbius addition and exponential/logarithmic maps.\"},{\"question\":\"How does the dissertation improve learning efficiency in low-label settings?\",\"answer\":\"It presents deep Bayesian active learning, using selection criteria such as BALanCe-based acquisition functions and strategies that support batch-mode learning under small or large batch sizes.\"}]","Machine Learning for Histopathology Images in Low-Data Regimes - Doctor of Philosophy Dissertation | PDF",1785681469,456,{"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-for-histopathology-images-in-low-data-regimes-doctor-of-philosophy-dissertation","",{"@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-for-histopathology-images-in-low-data-regimes-doctor-of-philosophy-dissertation/118089/",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 problem does the dissertation address about histopathology images?","Question",{"text":75,"@type":76},"It focuses on machine learning for histopathology image understanding when labeled data is limited, targeting multiple tasks including caption generation and instance-level classification with set-level labels.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"Which model is introduced to leverage non-Euclidean geometry?",{"text":80,"@type":76},"The dissertation proposes a hyperbolic attention model for histopathology images, built using Poincaré ball geometry concepts such as Möbius addition and exponential/logarithmic maps.",{"name":82,"@type":73,"acceptedAnswer":83},"How does the dissertation improve learning efficiency in low-label settings?",{"text":84,"@type":76},"It presents deep Bayesian active learning, using selection criteria such as BALanCe-based acquisition functions and strategies that support batch-mode learning under small or large batch sizes.","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"]