[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-123040-en":3,"doc-seo-123040-105":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},123040,8796095461564,"Liam","https://ap-avatar.wpscdn.com/davatar_155a257f0dc6eb9ab79c44ca47cae57d",8,"Research & Report","Algorithms and Analysis for Optimizing Robust Objectives in Fair Machine Learning","John Rawls’ veil of ignorance argument motivates fair justice and welfare judgments that treat everyone impartially, which this work formalizes as an adversarial min–max game. A daemon constructs a world and an angel selects placements to induce robustness, producing egalitarian-like maximin objectives. Weakening adversarial strength yields robust proxies for fair learning and allocation, deriving utilitarian, Gini, and power-mean welfare as special cases, plus a fused nonlinearity-based fairness concept. Robust fairness objectives can be efficiently optimized and supported by generalization bounds in machine learning.","arXiv :2404 .06703v1 [ cs .GT] 10 Apr 2024  \nAlgorithms and Analysis for Optimizing Robust Objectives  \nin Fair Machine Learning  \nCyrus Cousins  \nUniversity of Massachusetts Amherst  \nColumbia Workshop on Fairness in Operations and AI  \nDecember 2023  \nAbstract  \nThe original position or veil of ignorance argument of John Rawls, perhaps the most famous argument for egalitarianism, states that our concept of fairness, justice, or welfare should be decided from behind a veil of ignorance, and thus must consider everyone impartially (invariant to our identity) . This can be posedas a zero-sum game, where a Dæmon constructs a world, and an adversarial Angel then places the Dæmon into the world. This game incentivizes the Dæmon to maximize the minimum utility over all people (i.e., to maximize egalitarian welfare) . In some sense, this is the most extreme form of risk aversion or robustness, and we show that by weakening the Angel, milder robust objectives arise, which we argue are effective robust proxies for fair learning or allocation tasks. In particular, the utilitarian, Gini, and power-mean welfare concepts arise from special cases of the adversarial game, which has philosophical implications for the understanding of each of these concepts. We also motivate a new fairness concept that essentially fuses the nonlinearity of the power-mean with the piecewise nature of the Gini class. Then, exploiting the relationship between fairness and robustness, we show that these robust fairness concepts can all be efficiently optimized under mild conditions via standard maximin optimization techniques. Finally, we show that such methods apply in machine learning contexts, and moreover we show generalization bounds for robust fair machine learning tasks.  \nKeywords:  \nFair Machine Learning — Rawlsian Ethics — Adversarial Learning — Convex Optimization — Robust Fair  \nLearning  \n1 Introduction  \nFairness and robustness are crucial aspects of machine learning and allocation systems, both of which are generally addressed through modelling, data collection, and objective selection. This work extends ideas and objectives in welfare-centric fair machine learning and optimization introduced by Cousins [2021a,b, 2022, 2023] . We derive robust variants of fair objectives, and explore mathematical and philosophical connections between robustness and fairness. In particular, we consider robust welfare functions, which aggregate utility across a population, and robust malfare functions, which aggregate disutility, both serving as fairness metrics and as optimization targets. We then combine these robust objectives with adversarial optimization theory and techniques, which expandson the relationship between fairness, robustness, and uncertainty in machine learning and allocation problems [Mazzetto et al., 2021, Dong and Cousins, 2022, Cousins et al., 2023a] .  \nThe core of this paper is the construction of a hierarchy of Rawlsian games, where a Dæmon is tasked with creating a world, and an Angel places them within it. We consider various modifications and restrictions of this basic setup by adjusting the action space of the agents, as well as the payoff function, and show that various game theoretic solution concepts, including adversarial play for constant sum games and Nash equilibria for general sum games, give rise to various welfare and malfare concepts. Of course, this game is a metaphor, but it is strongly motivated by the grounded social planner’s problem, wherein a social planner seeks to organize society in a way that is favorable to all, and from these games we derive insight as to how the social planner should behave. The goals of this paper and the purpose of constructing this game are manifold.  \n1) We provide philosophical insight into a large class of welfare and malfare functions. Section 4.1 draws connections between fairness and robustness, finding that many classical welfare functions can be understood as robust utilitarian welfare ","cbCailgPCem2dXot","https://ap.wps.com/l/cbCailgPCem2dXot","pdf",2154949,1,17,"English","en",105,"# Introduction\n## Robust fairness and adversarial Rawlsian games\n## Contributions and roadmap\n## Optimization and generalization results","[{\"question\":\"How does the paper relate Rawls’ veil of ignorance to fair machine learning objectives?\",\"answer\":\"It translates the veil of ignorance into an adversarial min–max game that models fairness as an optimization problem over population utilities and disutilities. The resulting “robust” objectives capture egalitarian fairness under uncertainty.\"},{\"question\":\"What robust objectives does the paper derive from the adversarial game?\",\"answer\":\"The analysis shows that utilitarian, Gini, and power-mean welfare notions arise as special cases of the adversarial framework. It also motivates a new fairness concept combining power-mean nonlinearity with the piecewise structure of the Gini class.\"},{\"question\":\"How are the proposed robust fairness concepts optimized in machine learning and allocation settings?\",\"answer\":\"The paper leverages connections between robustness and fairness to show that robust fairness objectives can be efficiently optimized using standard maximin techniques. Under mild conditions the robust optimization attains a convex–concave structure suitable for first-order methods, with special cases reducing to linear or quadratic programming.\"}]","Algorithms and Analysis for Optimizing Robust Objectives in Fair Machine Learning | PDF",1785814347,43,{"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},"algorithms-and-analysis-for-optimizing-robust-objectives-in-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/algorithms-and-analysis-for-optimizing-robust-objectives-in-fair-machine-learning/123040/",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},"How does the paper relate Rawls’ veil of ignorance to fair machine learning objectives?","Question",{"text":75,"@type":76},"It translates the veil of ignorance into an adversarial min–max game that models fairness as an optimization problem over population utilities and disutilities. The resulting “robust” objectives capture egalitarian fairness under uncertainty.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"What robust objectives does the paper derive from the adversarial game?",{"text":80,"@type":76},"The analysis shows that utilitarian, Gini, and power-mean welfare notions arise as special cases of the adversarial framework. It also motivates a new fairness concept combining power-mean nonlinearity with the piecewise structure of the Gini class.",{"name":82,"@type":73,"acceptedAnswer":83},"How are the proposed robust fairness concepts optimized in machine learning and allocation settings?",{"text":84,"@type":76},"The paper leverages connections between robustness and fairness to show that robust fairness objectives can be efficiently optimized using standard maximin techniques. Under mild conditions the robust optimization attains a convex–concave structure suitable for first-order methods, with special cases reducing to linear or quadratic programming.","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"]