[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-123779-en":3,"doc-seo-123779-105":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},123779,3848291630094,"Emma Wilson","https://eur-avatar.wpscdn.com/davatar_085a072bc5b1113ac321206ff7593b45",8,"Research & Report","Robust Interactions with Machine Learning Models - Doctor of Philosophy Dissertation","This dissertation develops robust methods for interpreting and interacting with machine learning models under adversarial pressure. It analyzes how local, post hoc explanation methods can be manipulated through adversarial attacks and proposes evaluation protocols that test perturbation detection accuracy and hyperparameter sensitivity. It introduces reliable local explanations via Bayesian local modeling and presents natural-language systems, including TalkToModel and TABLET, to support developers in explainability dialogues and tabular prediction using instruction-based learning.","UC Irvine  \nUC Irvine Electronic Theses and Dissertations  \nTitle  \nRobust Interactions with Machine Learning Models  \nPermalink  \n[https://escholarship.org/uc/item/1gb45688](https://escholarship.org/uc/item/1gb45688)  \nAuthor  \nSlack, Dylan  \nPublication Date  \n2023  \nCopyright Information  \nThis work is made available under the terms of a Creative Commons Attribution License, availalbe 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  \nRobust Interactions with Machine Learning Models  \nDISSERTATION  \nsubmitted in partial satisfaction of the requirements  \nfor the degree of  \nDOCTOR OF PHILOSOPHY  \nin Computer Science  \nby  \nDylan Slack  \nDissertation Committee:  \nAssociate Professor Sameer Singh, Chair Assistant Professor Himabindu Lakkaraju Professor Erik Sudderth  \nChancellor’s Professor Padhraic Smyth  \n© 2023 Dylan Slack  \nDEDICATION  \nTo my family.  \nTABLE OF CONTENTS  \nPage  \nLIST OF FIGURES vi  \nLIST OF TABLES x  \nLIST OF ALGORITHMS xii  \nACKNOWLEDGMENTS xiii  \nVITA xv  \nABSTRACT OF THE DISSERTATION xviii  \n1 Introduction 1  \n1.1 Contributions & Key Findings .......................... 3  \n1.1.1 Robust Explainability .......................... 3  \n1.1.2 Conversational Systems for ML Development .............. 5  \n1.2 Declaration of Previous Work and Collaborations ............... 6  \nI Robust Machine Learning Explanations 8  \n2 Explainability Background 9  \n2.1 Notation ...................................... 10  \n2.2 Local Model Agnostic Explanations ....................... 10  \n2.3 Counterfactual Explanations ........................... 12  \n3 Adversarial Attacks On Local Post Hoc Explanations 16  \n3.1 Proposed Framework ............................... 17  \n3.2 Experimental Results ............................... 20  \n3.2.1 Effectiveness of Adversarial Classifiers ................. 23  \n3.2.2 Effect of Perturbation Detection Accuracy ............... 26  \n3.2.3 Robustness to Hyperparameters ..................... 27  \n3.2.4 Example Explanations .......................... 28  \n3.3 Summary of Contributions ............................ 28  \n4 Adversarial Attacks on Counterfactual Explanations 30  \n4.1 Adversarial Models for Manipulating Counterfactual Explanations ...... 33  \n4.2 Experiment Setup ................................. 38  \n4.3 Experiments .................................... 40  \n4.3.1 Effectiveness of the Manipulation .................... 41  \n4.3.2 Outlier Factor of Counterfactuals .................... 42  \n4.3.3 Potential Mitigation Strategies ...................... 43  \n4.4 Potential Impacts ................................. 45  \n4.5 Discussion ..................................... 46  \n4.6 Summary of Contribution ............................ 47  \n5 Reliable Local Post Hoc Explanations 48  \n5.1 Our Framework: Bayesian Local Explanations ................. 50  \n5.1.1 Constructing Bayesian Local Explanations ............... 51  \n5.1.2 Estimating the Number of Perturbations ................ 55  \n5.1.3 Focused Sampling of Perturbations ................... 57  \n5.2 Experiments .................................... 59  \n5.3 Discussion ..................................... 65  \n5.4 Summary of Contributions ............................ 65  \nII Natural Language For Explaining & Developing Machine Learning Models 67  \n6 Natural Language Conversations For Explainability With TalkToModel 68  \n6.1 User Study .................................... 69  \n6.1.1 Format ................................... 70  \n6.1.2 Results ................................... 71  \n6.1.3 Explainability Dialogue Desiderata from Interviewees ......... 73  \n6.2 TalkToModel ................................... 74  \n6.3 Methods ...................................... 77  \n6.3.1 Text Understandi","cbCaip29rVqpAWtA","https://ap.wps.com/l/cbCaip29rVqpAWtA","pdf",5482920,1,164,"English","en",105,"# Introduction\n## Contributions & Key Findings\n# Robust Machine Learning Explanations\n## Explainability Background\n## Adversarial Attacks On Local Post Hoc Explanations\n## Adversarial Attacks on Counterfactual Explanations\n## Reliable Local Post Hoc Explanations\n# Natural Language For Explaining & Developing Machine Learning Models\n## Natural Language Conversations For Explainability With TalkToModel\n## TABLET: Natural Language Instructions for Tabular Prediction\n# Conclusions and Future Work","[{\"question\":\"What key problem does this dissertation address?\",\"answer\":\"It addresses the robustness of interactions with machine learning models, focusing on how explanation methods can fail under adversarial attacks and how to build more reliable explanation and interaction mechanisms.\"},{\"question\":\"How does the work evaluate adversarial attacks on local explanations?\",\"answer\":\"It studies the effectiveness of adversarial classifiers, perturbation detection accuracy, robustness to hyperparameters, and provides example explanations to analyze attack impact on local post hoc methods.\"},{\"question\":\"What solutions are proposed for more reliable explanations and developer interaction?\",\"answer\":\"It proposes Bayesian local explanations for reliable local post hoc interpretation and introduces natural-language systems—TalkToModel for explainability dialogues and TABLET for instruction-driven tabular prediction.\"}]","Robust Interactions with Machine Learning Models - Doctor of Philosophy Dissertation | PDF",1785818526,413,{"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},"robust-interactions-with-machine-learning-models-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/robust-interactions-with-machine-learning-models-doctor-of-philosophy-dissertation/123779/",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 key problem does this dissertation address?","Question",{"text":75,"@type":76},"It addresses the robustness of interactions with machine learning models, focusing on how explanation methods can fail under adversarial attacks and how to build more reliable explanation and interaction mechanisms.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How does the work evaluate adversarial attacks on local explanations?",{"text":80,"@type":76},"It studies the effectiveness of adversarial classifiers, perturbation detection accuracy, robustness to hyperparameters, and provides example explanations to analyze attack impact on local post hoc methods.",{"name":82,"@type":73,"acceptedAnswer":83},"What solutions are proposed for more reliable explanations and developer interaction?",{"text":84,"@type":76},"It proposes Bayesian local explanations for reliable local post hoc interpretation and introduces natural-language systems—TalkToModel for explainability dialogues and TABLET for instruction-driven tabular prediction.","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"]