[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-121798-en":3,"doc-seo-121798-105":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},121798,16904993612988,"Olivia Brown","https://ap-avatar.wpscdn.com/davatar_a8503ba1806abce46bf441b54a3ca4cd",8,"Research & Report","Towards Robust and Fair Machine Learning","Towards Robust and Fair Machine Learning presents a research dissertation on improving both adversarial robustness and fairness in machine learning systems. The work studies how adversarial and poisoning strategies can degrade clustering and social robustness, and evaluates attacks against classical and deep clustering models, including a production-level MLaaS service. It then develops fairness-oriented clustering and video summarization methods using antidote data, and analyzes the interaction between adversarial robustness and social robustness through robust fair clustering and influence-based data selection.","UC Davis  \nUC Davis Electronic Theses and Dissertations  \nTitle  \nTowards Robust and Fair Machine Learning  \nPermalink  \n[https://escholarship.org/uc/item/9403w3j0](https://escholarship.org/uc/item/9403w3j0)  \nAuthor  \nChhabra, Anshuman  \nPublication Date  \n2023  \nPeer reviewed|Thesis/dissertation  \n[eScholarship.org](eScholarship.org) Powered by the California Digital Library  \nUniversity of California  \nTowards Robust and Fair Machine Learning  \nBy  \nANSHUMAN CHHABRA  \nDISSERTATION  \nSubmitted in partial satisfaction of the requirements for the degree of DOCTOR OF PHILOSOPHY  \nin  \nCOMPUTER SCIENCE  \nin the  \nOFFICE OF GRADUATE STUDIES  \nof the  \nUNIVERSITY OF CALIFORNIA  \nDAVIS  \nApproved:  \n\n| Prof. Prasant Mohapatra, Chair |\n| --- |\n| Prof. Xin Liu |\n\nProf. Hao-Chuan Wang Committee in Charge 2023  \n© Anshuman Chhabra, 2023 . All rights reserved.  \nTo my parents, Anjolie and Navdeep Chhabra.  \nii  \nContents  \nList of Figures v  \nList of Tables ix  \nAbstract xi  \nAcknowledgments xii  \nChapter 1 . Introduction 1  \n1.1. Motivation 1  \n1.2. Related Works 4  \n1.3. Contributions 6  \nChapter 2 . Analyzing the Adversarial Robustness of Models 8  \n2.1. Introduction 8  \n2.2. Poisoning Attacks Against Classical Clustering Models 9  \n2.3. Evasion Attacks Against Deep Clustering Models 19  \n2.4. Attacking Face++: A Production-Level MLaaS Clustering Service 31  \nChapter 3 . Fairness and Social Robustness in the Context of Machine Learning 33  \n3.1. Fair Clustering Using Antidote Data 33  \n3.2. Fair Video Summarization 43  \nChapter 4 . On the Interplay Between Adversarial and Social Robustness 61  \n4.1. Robust Fair Clustering 61  \n4.2. Enhancing Classifier Performance and Interpretability through Influence-Based Data Selection 79  \nChapter 5 . Conclusion 94  \nBibliography 95  \nList of Figures  \n2.1 Results on toy data after using Algorithm 1 for spill-over attack............ 15  \n2.2 Misclustered images that switched clusters from the Digit 4 to the Digit 1 cluster (adversarially perturbed sample shown with red border) ................. 17  \n2.3 Misclustered images that switched clusters from the Digit 9 to the Digit 8 cluster (adversarially perturbed sample shown with red border) ................. 17  \n2.4 Misclustered MNIST images that switched clusters from the Digit 4 to the Digit 1  \ncluster (adversarially perturbed sample shown with red border) ............. 18  \n2.5 Misclustered MNIST images that switched clusters from the Digit 3 to the Digit 2  \ncluster (adversarially perturbed sample shown with red border) ............. 18  \n2.6 Kama and Rosa wheat kernel clusters (target sample to be adversarially perturbed in red) visualized using the area, perimeter, and compactness features........... 19  \n2.7 Point1 and Grab pose clusters (target sample to be adversarially perturbed in red) visualized using the Z1 , Z2 , Z3 marker position features................. 20  \n2.8 Adversarial samples generated by our attack (first 4 image pairs from the left correspond to SPICE and the others to RUC) ......................... 23  \n2.9 Confusion matrices showcasing the effect of the attack for the SPICE/CC models on STL-10 ............................................ 28  \n2.10 Performance versus adversarial perturbation norm (STL-10) ............... 28  \n2.11 Transferability results showcasing post-attack (and pre-attack) NMI for different source/target models.................................... 29  \n2.12 Results on using anomaly detection as a possible defense................ 30  \n2.13 NMI/ACC/ARI before and after the attack on Face++ API............... 31  \n2.14 Adversarial samples for the Face++ attack......................... 32  \n3.1 Comparing clustering performance of Algorithm 3 with fair clustering algorithms  \nusing Silhouette scores. (Higher scores indicate better clustering performance. As  \ncan be observed, fair clusters obtained via Algorithm 3 achieve similar clustering performance to SOTA algorithms, while providing improved fairnes","cbCaier6GEJkXxBa","https://ap.wps.com/l/cbCaier6GEJkXxBa","pdf",23011473,1,134,"English","en",105,"# Contents\n## List of Figures\n## List of Tables\n## Abstract\n## Acknowledgments\n## Chapter 1. Introduction\n### 1.1 Motivation\n### 1.2 Related Works\n### 1.3 Contributions\n## Chapter 2. Analyzing the Adversarial Robustness of Models\n### 2.1 Introduction\n### 2.2 Poisoning Attacks Against Classical Clustering Models\n### 2.3 Evasion Attacks Against Deep Clustering Models\n### 2.4 Attacking Face++: A Production-Level MLaaS Clustering Service\n## Chapter 3. Fairness and Social Robustness in the Context of Machine Learning\n### 3.1 Fair Clustering Using Antidote Data\n### 3.2 Fair Video Summarization\n## Chapter 4. On the Interplay Between Adversarial and Social Robustness\n### 4.1 Robust Fair Clustering\n### 4.2 Enhancing Classifier Performance and Interpretability through Influence-Based Data Selection\n## Chapter 5. Conclusion\n## Bibliography","[{\"question\":\"What does the dissertation focus on regarding machine learning reliability and fairness?\",\"answer\":\"The dissertation focuses on adversarial robustness and fairness, examining how adversarial and poisoning attacks impact clustering and related services, and how fairness objectives can be incorporated into learning systems.\"},{\"question\":\"Which types of attacks are studied for clustering models?\",\"answer\":\"It studies poisoning attacks against classical clustering models and evasion attacks against deep clustering models, including attacks on a production-level MLaaS clustering service (Face++).\"},{\"question\":\"How does the work connect adversarial robustness with social robustness?\",\"answer\":\"It analyzes their interplay through robust fair clustering and methods that enhance classifier performance and interpretability using influence-based data selection.\"}]","Towards Robust and Fair Machine Learning | PDF",1785806931,338,{"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},"towards-robust-and-fair-machine-learning","",{"@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/towards-robust-and-fair-machine-learning/121798/",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 does the dissertation focus on regarding machine learning reliability and fairness?","Question",{"text":75,"@type":76},"The dissertation focuses on adversarial robustness and fairness, examining how adversarial and poisoning attacks impact clustering and related services, and how fairness objectives can be incorporated into learning systems.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"Which types of attacks are studied for clustering models?",{"text":80,"@type":76},"It studies poisoning attacks against classical clustering models and evasion attacks against deep clustering models, including attacks on a production-level MLaaS clustering service (Face++).",{"name":82,"@type":73,"acceptedAnswer":83},"How does the work connect adversarial robustness with social robustness?",{"text":84,"@type":76},"It analyzes their interplay through robust fair clustering and methods that enhance classifier performance and interpretability using influence-based data selection.","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"]