[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-123608-en":3,"doc-seo-123608-105":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":20,"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},123608,1099513958762,"Logic","https://ap-avatar.wpscdn.com/avatar/1000023916a998db790?x-image-process=image/resize,m_fixed,w_180,h_180&k=1784791008015729253",8,"Research & Report","A Survey on Intersectional Fairness in Machine Learning - Notions, Mitigation, and Challenges","Widespread deployment of machine learning in decision-critical domains like criminal sentencing and bank lending has intensified scrutiny of fairness failures and the discrimination they may encode. This survey reviews how algorithms and metrics are used to measure and mitigate bias, and highlights intersectional bias, where multiple sensitive attributes jointly shape distinct patterns of disadvantage. It proposes a taxonomy of intersectional fairness notions and mitigation strategies, then summarizes key research challenges and future directions.","A Survey on Intersectional Fairness in Machine Learning: Notions, Mitigation,  \nand Challenges  \nUsman Gohar 1 , Lu Cheng2  \n1Iowa State University  \n2University of Illinois Chicago  \n[ugohar@iastate.edu](ugohar@iastate.edu), [lucheng@uic.edu](lucheng@uic.edu)  \narXiv :2305 .06969v2 [ cs .LG] 12 May 2023  \nAbstract  \nThe widespread adoption of Machine Learning systems, especially in more decision-critical applications such as criminal sentencing and bank loans, has led to increased concerns about fairness implications. Algorithms and metrics have been developed to mitigate and measure these discriminations. More recently, works have identified a more challenging form of bias called intersectional bias, which encompasses multiple sensitive attributes, such as race and gender, together. In this survey, wereview the state-of-the-art in intersectional fairness.  \nWe present a taxonomy for intersectional notions of fairness and mitigation. Finally, we identify the key challenges and provide researchers with guidelines for future directions.  \n1 Introduction  \nMachine learning (ML) has been increasingly used in highstake applications such as loans, criminal sentencing, and hiring decisions with reported fairness implications for different demographic groups [Cheng et al., 2021] . Measuring and mitigating discrimination in ML/AI systems has been studied extensively [Mehrabi et al., 2021] . Such works have focused on two specific categories of algorithmic fairness: Group or individual fairness. The majority of early group fairness research was focused on one dimension of group identity, e.g., race or gender. This setting is defined as independent groups fairness [Yang et al., 2020a] . However, recent works have identified a more nuanced case of group unfairness that spans multiple subgroups based on Crenshaw's theory of “intersectionality” [Crenshaw, 1989] called intersectional group fairness. At a high level, intersectionality states that interaction along multiple dimensions of identity produces unique and differing levels of discrimination for various possible subgroups, e.g., a Black woman's experience of discrimination differs from both women and Black people in general. Finally, gerrymandering groups are the union of independent and intersectional groups. Figure 1 shows an example of these group fairness definitions using “gender” and “race”.  \nBy categorizing people only into distinct overlapping groups, independent group fairness fails to consider the discrimination people face at the intersection of such groups.  \nFigure 1: Definitions of group fairness [Yang et al., 2020a] .  \nThis has been well-studied in philosophy and social psychology (e.g., [Bierly, 1985; Akrami et al., 2011]), but recent works also demand urgency to do so in ML fairness. Specifically, an ML predictor might be fair w.r.t the independent groups but not intersectional groups. For example, [Buolamwini and Gebru, 2018] identified accuracy disparities that were more significant for Black Women in gender classification algorithms, compared to independent groups. In NLP, works have evaluated popular generative models [Kirk et al., 2021; Tan and Celis, 2019] and also identified such cases of intersectional bias.  \nCompared to the binary view of fairness in the independent case, the problem of intersectional fairness poses unique challenges. For instance, for what level of granularity of intersectional groups should fairness be guaranteed? On the other hand, smaller subgroups have higher data sparsity, resulting in higher uncertainty [Foulds et al., 2018] . Furthermore, an intersectional identity often amplifies biases that might not exist in its constituent groups (e.g., Black woman vs. Black or Woman), rendering traditional mitigation techniques ineffective. To this end, an emerging body of work, e.g., subgroup fairness [Kearns et al., 2018] and multicalibration [HebertJohnson et al., 2018], has proposed various notions of intersectional fairness and mitigation technique","cbCaiuqOcFSXHdZk","https://ap.wps.com/l/cbCaiuqOcFSXHdZk","pdf",416638,1,9,"English","en",105,"# Introduction\n## Background on ML fairness and independent group fairness\n## Intersectional group fairness and intersectionality\n## Challenges unique to intersectional fairness\n# Notions of Intersectional Fairness\n## Intersectionality and the limits of traditional group fairness\n## Subgroup complexity and balancing group vs individual fairness","[{\"question\":\"What problem does intersectional fairness address that independent group fairness misses?\",\"answer\":\"Independent group fairness evaluates discrimination across separate protected attributes, but it can overlook disadvantages arising from the combined intersection of attributes, such as race and gender simultaneously.\"},{\"question\":\"Why are traditional fairness guarantees difficult to apply to intersectionality?\",\"answer\":\"Intersectionality can involve potentially infinite or very large numbers of overlapping subgroups, making it infeasible to satisfy traditional group fairness requirements across all intersections.\"},{\"question\":\"What does the survey contribute to the study of intersectional bias in ML?\",\"answer\":\"It presents a taxonomy of intersectional fairness notions and fair learning methods, analyzes representative approaches and their limitations, and concludes with key challenges and open problems for future research.\"}]","A Survey on Intersectional Fairness in Machine Learning - Notions, Mitigation, and Challenges | PDF",1785817610,23,{"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},"a-survey-on-intersectional-fairness-in-machine-learning-notions-mitigation-and-challenges","",{"@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/a-survey-on-intersectional-fairness-in-machine-learning-notions-mitigation-and-challenges/123608/",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":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"What problem does intersectional fairness address that independent group fairness misses?","Question",{"text":75,"@type":76},"Independent group fairness evaluates discrimination across separate protected attributes, but it can overlook disadvantages arising from the combined intersection of attributes, such as race and gender simultaneously.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"Why are traditional fairness guarantees difficult to apply to intersectionality?",{"text":80,"@type":76},"Intersectionality can involve potentially infinite or very large numbers of overlapping subgroups, making it infeasible to satisfy traditional group fairness requirements across all intersections.",{"name":82,"@type":73,"acceptedAnswer":83},"What does the survey contribute to the study of intersectional bias in ML?",{"text":84,"@type":76},"It presents a taxonomy of intersectional fairness notions and fair learning methods, analyzes representative approaches and their limitations, and concludes with key challenges and open problems for future research.","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},[93,97,101,105,110,115,120,123,127,130,134],{"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":21,"doc_module":4,"doc_module_name":46,"category_name":124,"show_sort_weight":125,"slug":126},"Religion & Spirituality",20,"religion-spirituality",{"id":125,"doc_module":4,"doc_module_name":46,"category_name":128,"show_sort_weight":125,"slug":129},"World Cup","world-cup",{"id":131,"doc_module":4,"doc_module_name":46,"category_name":132,"show_sort_weight":131,"slug":133},10,"Lifestyle","lifestyle",{"id":135,"doc_module":4,"doc_module_name":46,"category_name":136,"show_sort_weight":106,"slug":137},19,"General","general"]