[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-125619-en":3,"doc-seo-125619-105":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},125619,549758146520,"Patrick","https://ap-avatar.wpscdn.com/avatar/80002397d8c0411e94?_k=1775819394049821470",8,"Research & Report","Causal Fair Machine Learning via Rank-Preserving Interventional Distributions","The work develops a causal notion of fairness for automated decision-making models that mitigate unfairness tied to protected attributes. It defines individuals as normatively equal in a fictitious FiND world where protected attributes have no direct or indirect causal effect on the target. Rank-preserving interventional distributions are introduced to construct the corresponding estimand, along with a warping method to approximate the FiND world from observed data. Evaluation uses simulations and empirical evidence, showing the approach can identify most discriminated individuals and reduce unfairness.","arXiv :2307 . 12797v 1 [ cs .LG] 24 Jul 2023  \nCausal Fair Machine Learning via Rank-Preserving Interventional Distributions  \nLudwig Bothmann1,2, * , Susanne Dandl 1,2 and Michael Schomaker1,3  \n1 Department of Statistics, LMU Munich, Germany 2 Munich Center for Machine Learning (MCML)  \n3 Center for Infectious Disease Epidemiology, School of Public Health, University of Cape Town, South Africa  \nAbstract  \nA decision can be defined as fair if equal individuals are treated equally and unequals unequally. Adopting this definition, the task of designing machine learning models that mitigate unfairness in automated decision-making systems must include causal thinking when introducing protected attributes. Following a recent proposal, we define individuals as being normatively equal if they are equal in a fictitious, normatively desired (FiND) world, where the protected attribute has no (direct or indirect) causal effect on the target. We propose rank-preserving interventional distributions to define an estimand of this FiND world and a warping method for estimation. Evaluation criteria for both the method and resulting model are presented and validated through simulations and empirical data. With this, we show that our warping approach effectively identifies the most discriminated individuals and mitigates unfairness.  \nKeywords  \nfairness in ML, causal thinking, interventional distributions, stochastic interventions, rank-preserving interventions, quasi-individual fairness  \n1. Introduction  \nAutomated decision-making (ADM) systems can support human decision-makers by predicting some variable of interest via a machine learning (ML) model. The data used for learning such ML models can have historical bias, i.e., show normatively undesirable discrimination against certain groups of protected attributes (PAs) . When left unaddressed, this historical bias leads to biased ML models, generating fairness problems in ADM systems. The research field of fair machine learning (fairML) has quickly grown around this problem in recent years, giving birth to various “fairness metrics”(such as, e.g., demographical parity, see [1] for an overview) .  \nA critique raised by [2] is that the question of what fairness is – as a philosophical concept – is rarely discussed. Hence, the proposed fairness metrics lack a philosophical justification, making it unclear which concept of fairness is measured by the respective metrics. They propose a consistent concept of fairness and outline how this should be integrated into the design of  \nAEQUITAS 2023 – Workshop on Fairness and Bias in AI at 26th European Conference on Artificial Intelligence  \n* Corresponding author.  \n\" [ludwig.bothmann@stat.uni-muenchen.de](ludwig.bothmann@stat.uni-muenchen.de) (L. Bothmann); [susanne.dandl@stat.uni-muenchen.de](susanne.dandl@stat.uni-muenchen.de) (S. Dandl);  \n[michael.schomaker@stat.uni-muenchen.de](michael.schomaker@stat.uni-muenchen.de) (M. Schomaker)  \n~ [https://www.slds.stat.uni-muenchen.de/people/bothmann/](https://www.slds.stat.uni-muenchen.de/people/bothmann/) (L. Bothmann)  \n􀀒 0000-0002-1471-6582 (L. Bothmann); 0000-0003-4324-4163 (S. Dandl); 0000-0002-8475-0591 (M. Schomaker)  \n © 2023 Copyright for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0) . CWPEURorkroceshopedings http://ceurISSN 1613-ws-0073.org CEUR Workshop Proceedings ([CEUR-WS.org](CEUR-WS.org))  \nML models in ADM systems. Following Aristotle [3], they define a treatment as being fair “if equals are treated equally and if unequals are treated unequally”. Furthermore, they distinguish between descriptively unfair treatment (which can occur without PAs) and normatively unfair treatment (which is a causal notion) . For this, they conceive a fictitious, normatively desired (FiND) world, where the PA has no (direct or indirect) causal effect on the target variable. Individuals are normatively considered equal if they are equal in the ","cbCaial6bNjo7QCH","https://ap.wps.com/l/cbCaial6bNjo7QCH","pdf",963165,1,17,"English","en",105,"# Introduction\n# Related Work","[{\"question\":\"What does the paper mean by “causal fairness” in machine learning?\",\"answer\":\"Fairness is defined using a causal perspective: individuals are considered normatively equal in a FiND world where protected attributes have no direct or indirect causal effect on the target. This connects fairness directly to causal effects rather than only statistical group differences.\"},{\"question\":\"How are rank-preserving interventional distributions used?\",\"answer\":\"They define an estimand for the FiND world by applying specific stochastic interventions that remove causal paths from the protected attribute to the target. The interventions are designed to be rank-preserving so disadvantaged individuals keep their group-specific real-world rank as FiND-world population ranks.\"},{\"question\":\"How does the proposed warping method help in practice?\",\"answer\":\"The warping method maps real-world data into a warped representation that approximates the FiND world. An ML model trained on this warped data can then be used at prediction time after warping new observations, reducing the influence of protected attributes.\"}]","Causal Fair Machine Learning via Rank-Preserving Interventional Distributions | PDF",1785900257,43,{"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},"causal-fair-machine-learning-via-rank-preserving-interventional-distributions","",{"@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/causal-fair-machine-learning-via-rank-preserving-interventional-distributions/125619/",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","2026-08-05",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 paper mean by “causal fairness” in machine learning?","Question",{"text":75,"@type":76},"Fairness is defined using a causal perspective: individuals are considered normatively equal in a FiND world where protected attributes have no direct or indirect causal effect on the target. This connects fairness directly to causal effects rather than only statistical group differences.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How are rank-preserving interventional distributions used?",{"text":80,"@type":76},"They define an estimand for the FiND world by applying specific stochastic interventions that remove causal paths from the protected attribute to the target. The interventions are designed to be rank-preserving so disadvantaged individuals keep their group-specific real-world rank as FiND-world population ranks.",{"name":82,"@type":73,"acceptedAnswer":83},"How does the proposed warping method help in practice?",{"text":84,"@type":76},"The warping method maps real-world data into a warped representation that approximates the FiND world. An ML model trained on this warped data can then be used at prediction time after warping new observations, reducing the influence of protected attributes.","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"]