[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-123421-en":3,"doc-seo-123421-105":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},123421,13056703019662,"Evangeline","https://ap-avatar.wpscdn.com/avatar/be000253a8e92610077?_k=1778726343310543188",8,"Research & Report","Interpretable Machine Learning and Trustworthy Uncertainty Quantification - Dissertation","As machine learning models increasingly support high-stakes decisions, understanding their reasoning and managing the risks behind those decisions becomes essential. This dissertation develops methods that combine interpretable machine learning with trustworthy uncertainty quantification, motivated by interdisciplinary needs in clinical medicine and genomics. It introduces interpretable approaches for tree-based models, scalable interaction attribution for large language models, and a PCS-based framework that supports decision-making and calibrated, adaptive prediction sets for reliable model-assisted clinical actions.","UC Berkeley  \nUC Berkeley Electronic Theses and Dissertations  \nTitle  \nInterpretable Machine Learning and Trustworthy Uncertainty Quantification  \nPermalink  \n[https://escholarship.org/uc/item/2sr1c8nk](https://escholarship.org/uc/item/2sr1c8nk)  \nISBN  \n9798293892792  \nAuthor  \nAgarwal, Abhineet  \nPublication Date  \n2025-08-01  \nPeer reviewed|Thesis/dissertation  \n[eScholarship.org](eScholarship.org) Powered by the California Digital Library  \nUniversity of California  \nInterpretable Machine Learning and Trustworthy Uncertainty Quantification  \nby  \nAbhineet Agarwal  \nA dissertation submitted in partial satisfaction of the requirements for the degree of  \nDoctor of Philosophy  \nin  \nStatistics  \nin the  \nGraduate Division  \nof the  \nUniversity of California, Berkeley  \nCommittee in charge:  \nProfessor Bin Yu, Chair  \nAssociate Professor Peng Ding  \nAssistant Professor Sam Pimentel  \nSummer 2025  \nInterpretable Machine Learning and Trustworthy Uncertainty Quantification  \nCopyright 2025  \nby  \nAbhineet Agarwal  \n1  \nAbstract  \nInterpretable Machine Learning and Trustworthy Uncertainty Quantification  \nby  \nAbhineet Agarwal  \nDoctor of Philosophy in Statistics  \nUniversity of California, Berkeley  \nProfessor Bin Yu, Chair  \nAs machine learning (ML) models are increasingly integrated into high-stakes domains, understanding their decision-making processes and the risks those decisions entail is critical. This dissertation advances methods for interpretable ML and trustworthy uncertainty quantification (UQ), motivated by interdisciplinary challenges in clinical medicine and genomics. The dissertation is organized into three parts. Part I focuses on interpretability for tree-based models—an important and inherently interpretable class of models widely used in clinical decision-making and the detection of epistatic genetic interactions. We first introduce Fast Interpretable Greedy-Tree Sums (FIGS), a tree-based algorithm that significantly improves predictive performance while retaining interpretability. We then propose MDI+, a feature importance method for Random Forests grounded in the Predictability–Computability–Stability (PCS) framework for veridical data science, yielding more reliable estimates of feature importance and interactions. Part II turns to interpretability in large-scale language models (LLMs) . We develop scalable algorithms for interaction attribution, drawing on tools from signal processing and information theory. These methods enable the detection of higher-order feature interactions in long-context settings that were previously out of reach. Part III addresses decision-making and trustworthy UQ through the PCS framework. We first combine PCS with LLMs to improve clinical lab test selection in emergency departments. Finally, we present a PCS-guided method for constructing calibrated and adaptive prediction sets, enabling more reliable model-assisted decisions in high-stakes clinical applications.  \ni  \nContents  \nContents i  \n1 Overview 1  \nI Improving Interpretability for Tree-Based Models 5  \n2 Fast Interpretable Greedy Tree Sums 6  \n2.1 Introduction ...................................... 6  \n2.2 FIGS: Algorithm description and run-time ...................... 9  \n2.3 Related work and its connections to FIGS ...................... 11  \n2.4 FIGS results on real-world benchmark datasets ................... 12  \n2.5 Learning CDIs via FIGS ............................... 13  \n2.6 Theoretical Investigations ............................... 18  \n2.7 Bagging-FIGS ..................................... 19  \n2.8 Discussion ....................................... 20  \n3 Integrating RFs and GLMs 22  \n3.1 Introduction ...................................... 22  \n3.2 Related Work ..................................... 24  \n3.3 A Linear Regression Perspective of Decision Trees ................. 26  \n3.4 Introducing RF+ and MDI+ .............................. 30  \n3.5 RF+ Prediction Performance ............................. 34 ","cbCaigukvjMprNTl","https://ap.wps.com/l/cbCaigukvjMprNTl","pdf",5985277,1,204,"English","en",105,"# Overview\n## Improving Interpretability for Tree-Based Models\n## Fast Interpretable Greedy Tree Sums\n## Integrating RFs and GLMs\n# Scalable Interaction Detection for Large Language Models\n## SPEX: Scaling Feature Interaction Explanations for LLMs\n# Enhancing Clinical-Decision Making and Trustworthy Uncertainty Quantification\n## ED-Copilot\n## PCS-UQ","[{\"question\":\"What is the core focus of the dissertation?\",\"answer\":\"The dissertation focuses on interpretable machine learning and trustworthy uncertainty quantification, aimed at understanding model decisions and making them safer in high-stakes domains.\"},{\"question\":\"Which methods are proposed for tree-based models?\",\"answer\":\"It proposes FIGS for tree-based interpretability with improved predictive performance, and MDI+ for more reliable feature importance and interaction estimates in Random Forests.\"},{\"question\":\"How does the dissertation address uncertainty and decision-making?\",\"answer\":\"It uses the PCS framework to guide calibrated and adaptive prediction sets, including methods that improve clinical lab test selection and support more reliable model-assisted clinical decisions.\"}]","Interpretable Machine Learning and Trustworthy Uncertainty Quantification - Dissertation | PDF",1785816379,514,{"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},"interpretable-machine-learning-and-trustworthy-uncertainty-quantification-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/interpretable-machine-learning-and-trustworthy-uncertainty-quantification-dissertation/123421/",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","2026-08-04",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 is the core focus of the dissertation?","Question",{"text":75,"@type":76},"The dissertation focuses on interpretable machine learning and trustworthy uncertainty quantification, aimed at understanding model decisions and making them safer in high-stakes domains.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"Which methods are proposed for tree-based models?",{"text":80,"@type":76},"It proposes FIGS for tree-based interpretability with improved predictive performance, and MDI+ for more reliable feature importance and interaction estimates in Random Forests.",{"name":82,"@type":73,"acceptedAnswer":83},"How does the dissertation address uncertainty and decision-making?",{"text":84,"@type":76},"It uses the PCS framework to guide calibrated and adaptive prediction sets, including methods that improve clinical lab test selection and support more reliable model-assisted clinical decisions.","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"]