[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-117423-en":3,"doc-seo-117423-105":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},117423,13056703019404,"Miles","https://ap-avatar.wpscdn.com/davatar_29158cc5080c5b710cf443261637dec0",8,"Research & Report","Trustworthy Machine Learning under Social and Adversarial Data Sources - Thesis Abstract","Machine learning research explores how social and adversarial interactions affect learning, prediction, and decision-making performance. It studies scenarios where data is generated by strategic individuals, collected by self-interested data collectors, corrupted by adversarial attackers, and used to form predictors, models, and policies with multiple objectives. The work addresses degradation such as vulnerability of deep neural networks to adversarial examples and reduced performance of classical algorithms under strategic behavior. The thesis organizes results into two parts: learning under social data sources and adversarial data sources, including robust learning, incentives in federated settings, and transformation-invariance via data augmentation.","arXiv :2408 .01596v1 [ cs .LG] 2 Aug 2024  \nTrustworthy Machine Learning under Social and Adversarial Data Sources  \nBY  \nHAN SHAO  \nA thesis submitted  \nin partial fulfillment of the requirements for  \nthe degree of  \nDoctor of Philosophy in Computer Science  \nat the  \nTOYOTA TECHNOLOGICAL INSTITUTE AT CHICAGO Chicago, Illinois  \nJuly, 2024  \nThesis Committee:  \nAvrim Blum (Thesis Advisor)  \nNika Haghtalab  \nYishay Mansour  \nNati Srebro  \nAbstract  \nMachine learning has witnessed remarkable breakthroughs in recent years. As machine learning permeates various aspects of daily life, individuals and organizations increasingly interact with these systems, exhibiting a wide range of social and adversarial behaviors. These behaviors may have a notable impact on the behavior and performance of machine learning systems. Specifically, during these interactions, data may be generated by strategic individuals, collected by self-interested data collectors, possibly poisoned by adversarial attackers, and used to create predictors, models, and policies satisfying multiple objectives. As a result, the machine learning systems’ outputs might degrade, such as the susceptibility of deep neural networks to adversarial examples (Shafahi et al., 2018; Szegedy et al., 2013) and the diminished performance of classic algorithms in the presence of strategic individuals (Ahmadi et al., 2021) . Addressing these challenges is imperative for the success of machine learning in societal settings.  \nThis thesis is organized into two parts: learning under social data sources and adversarial data sources. For social data sources, we consider problems including: (1) learning with strategic individuals for both finite and infinite hypothesis classes, where we provide an understanding of learnability in both online and PAC strategic settings,(2) incentives and defections of selfinterested data collectors in single-round federated learning, multi-round federated learning, and collaborative active learning,(3) learning within games, in which one of players runs a learning algorithm instead of best responding, and (4) multi-objective learning in both decision making and online learning. For adversarial data sources, we study problems including: (1) robust learning under clean-label attacks, where the attacker injects a set of correctly labeled points into the training set to mislead the learner into making mistakes at targeted test points, and (2) learning under transformation invariances and analyzing the popular method of data augmentation.  \nAcknowledgements  \nI couldn’t have been more fortunate to be advised by the best advisor in the world, Avrim Blum. Avrim is an incredible mentor—wise, smart, knowledgeable, kind, supportive, and trustworthy. From him, I learned how to conduct research, develop my taste in problems, model and formulate questions, and find solutions. Avrim gave me the freedom to explore different problems, encouraged me whenever I got stuck, and helped me grow as an independent researcher. He provided valuable and insightful suggestions whenever I faced uncertainty in my decisions. Avrim is my role model as a researcher, advisor, and human being. Avrim, thankyou. I’ve truly enjoyed my PhD journey under your guidance and wouldn’t have come this far in academia without you. I aspire to be an advisor like you in the future, though I know it’s a really high bar to reach.  \nI learned a lot from working with Yishay Mansour and Shay Moran. They are both incredibly smart and supportive. I was lucky to have worked with them, especially during my job search season. The highlight of my week was always our meeting time, where I received enlightening research ideas, job search suggestions, and mental support. I was very stressed during that time, and things wouldn’t have gone as smoothly without your help.  \nI’d like to thank Aaron Roth for hosting me as a visiting student during the summer of 2023 . I had a wonderful time in Philly. Aaron is smart, suppor","cbCaivxGb4FzxIoN","https://ap.wps.com/l/cbCaivxGb4FzxIoN","pdf",5533222,1,376,"English","en",105,"# Abstract\n## Part I: Learning under Social Data Sources\n## Part II: Adversarial Data Sources\n# Acknowledgements","[{\"question\":\"What challenges does the thesis address in trustworthy machine learning?\",\"answer\":\"It focuses on social and adversarial behaviors that influence how training data is generated, collected, poisoned, and ultimately used to build models and policies. It highlights how these behaviors can degrade system outputs and reduce learning performance.\"},{\"question\":\"What topics are covered under learning with social data sources?\",\"answer\":\"The thesis covers learning with strategic individuals in finite and infinite hypothesis classes, incentives and defections of self-interested data collectors in federated and collaborative settings, learning within games where a player runs a learning algorithm, and multi-objective learning for decision making and online learning.\"},{\"question\":\"What adversarial learning problems does the thesis study?\",\"answer\":\"It studies robust learning under clean-label attacks, where correctly labeled points are injected to mislead the learner, and learning under transformation invariances by analyzing data augmentation methods.\"}]","Trustworthy Machine Learning under Social and Adversarial Data Sources - Thesis Abstract | PDF",1785675801,948,{"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},"trustworthy-machine-learning-under-social-and-adversarial-data-sources-thesis-abstract","",{"@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/trustworthy-machine-learning-under-social-and-adversarial-data-sources-thesis-abstract/117423/",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 challenges does the thesis address in trustworthy machine learning?","Question",{"text":75,"@type":76},"It focuses on social and adversarial behaviors that influence how training data is generated, collected, poisoned, and ultimately used to build models and policies. It highlights how these behaviors can degrade system outputs and reduce learning performance.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"What topics are covered under learning with social data sources?",{"text":80,"@type":76},"The thesis covers learning with strategic individuals in finite and infinite hypothesis classes, incentives and defections of self-interested data collectors in federated and collaborative settings, learning within games where a player runs a learning algorithm, and multi-objective learning for decision making and online learning.",{"name":82,"@type":73,"acceptedAnswer":83},"What adversarial learning problems does the thesis study?",{"text":84,"@type":76},"It studies robust learning under clean-label attacks, where correctly labeled points are injected to mislead the learner, and learning under transformation invariances by analyzing data augmentation methods.","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"]