[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-122432-en":3,"doc-seo-122432-105":30,"detail-sidebar-cat-0-en-105":83},{"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},122432,13056703019662,"Evangeline","https://ap-avatar.wpscdn.com/avatar/be000253a8e92610077?_k=1778726343310543188",8,"Research & Report","Scarce Resource Allocations That Rely On Machine Learning - Should Be Randomized","Contrary to deterministic views of algorithmic fairness, this paper argues that fairly allocating scarce resources through machine learning often requires randomness. It explains why, when, and how to randomize using stochastic procedures that better reflect individuals’ claims to social goods or opportunities. The work addresses bias amplification from deterministic scoring and thresholds, and proposes randomized allocation to better respect claims even when individuals do not receive the resource.","Scarce Resource Allocations That Rely On Machine Learning  \nShould Be Randomized  \nShomik Jain 1 Kathleen Creel 2 Ashia Wilson 1 3  \narXiv :2404 .08592v3 [ cs .CY] 19 Jun 2024  \nAbstract  \nContrary to traditional deterministic notions of algorithmic fairness, this paper argues that fairly allocating scarce resources using machine learning often requires randomness. We address why, when, and how to randomize by proposing stochastic procedures that more adequately account for all of the claims that individuals have to allocations of social goods or opportunities.  \n1. Introduction  \nSometimes resources or opportunities are scarce: jobs, welfare benefits, or life-saving medicines cannot be divided among all those who deserve them. Worse yet, it is often unclear which individuals are most deserving. Perhaps they all are. Decision-makers hope to use algorithmic systems to allocate scarce resources and goods fairly. But without careful attention, it is easy for algorithms to replicate or amplify the biases and inequalities in their training data.  \nThe fair machine learning community has developed sophisticated theoretical and formal tools to reduce algorithmic bias, increase fairness, and promote justice. However, these tools are almost exclusively deterministic. For example, employers with more qualified applicants than job openings often rely on hiring algorithms to screen applicants for interviews (Raghavan et al., 2020) . These algorithms assign a score or ranking to candidates. Employers then threshold these scores or rankings to deterministically pick candidates to interview. Similarly, healthcare providers often have a limited supply of life-saving medical resources such as ventilators, therapeutics, or organs. Patients are often triaged based on algorithms that predict their survival rate or life expectancy post-treatment (Chin et al., 2023) . Most existing  \n1Institute for Data, Systems, and Society, MIT 2Department of Philosophy & Religion and Khoury College of Computer Sciences, Northeastern University 3Department of Electrical Engineering and Computer Science, MIT. Correspondence to:\u003C[shomikj@mit.edu](shomikj@mit.edu) >.  \nProceedings of the 41 st International Conference on Machine Learning, Vienna, Austria. PMLR 235, 2024 . Copyright 2024 by the author(s) .  \nwork on algorithmic fairness relies on deterministic algorithms to incorporate fairness. Once algorithmic bias has been reduced to the extent possible, the algorithm allocates resources to the top candidate(s) . If Alice is the top-ranked candidate for every job or has the most expected qualityadjusted life-years, she should deterministically receive the job offer or organ every time.  \nRecent works on arbitrariness and fairness suggest that even counterfactual non-determinism can be unfair. If there exist many possible models with similar predictive performance but slightly different decisions on individuals, a state of affairs called “predictive multiplicity”(Marx et al., 2020) or “model multiplicity”(Black et al., 2022), it is unfair to naively pick one of the models for our decision-making algorithm (Hsu & Calmon, 2022) . Instead, we should reduce multiplicity by altering the training process to reduce the variance that leads to diverging predictions (Cooper et al., 2023), especially on under-represented individuals (Ganesh et al., 2023), iterate until predictions agree about individuals (Roth et al., 2023) or even abstain from making predictions on some people altogether (Cooper et al., 2023) .  \nWhile sharing the goal of reducing bias and increasing fairness, this work argues that the fair machine learning community has underutilized non-determinism and randomization as tools to achieve fairness. In some settings that involve algorithmic decision-making, we contend that nondeterminism is required for fair outcomes. In what follows, we first motivate why and when fairness requires randomization. We adopt philosopher John Broome’s concept of the value of lott","cbCaiesh0ob8HtKU","https://ap.wps.com/l/cbCaiesh0ob8HtKU","pdf",20146436,1,26,"English","en",105,"# Introduction\n## Related Work\n## Randomization Rationale and Motivation\n## Formalizing Randomization Methods","[{\"question\":\"Why might deterministic fairness methods be insufficient?\",\"answer\":\"Because predictions under uncertainty can lead to unfairly consistent commitments to mistakes, and repeated or multi-shot decisions can compound errors into patterned inequality, which randomness can help avoid.\"}]","Scarce Resource Allocations That Rely On Machine Learning - Should Be Randomized | PDF",1785810613,66,{"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":78,"head_meta":80,"extra_data":82,"updated_unix":28},"scarce-resource-allocations-that-rely-on-machine-learning-should-be-randomized","",{"@graph":36,"@context":77},[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/scarce-resource-allocations-that-rely-on-machine-learning-should-be-randomized/122432/",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],{"name":72,"@type":73,"acceptedAnswer":74},"Why might deterministic fairness methods be insufficient?","Question",{"text":75,"@type":76},"Because predictions under uncertainty can lead to unfairly consistent commitments to mistakes, and repeated or multi-shot decisions can compound errors into patterned inequality, which randomness can help avoid.","Answer","https://schema.org",{"og:url":52,"og:type":79,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":81,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":84},[85,89,93,97,102,107,112,115,120,123,127],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":86,"show_sort_weight":87,"slug":88},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":90,"show_sort_weight":91,"slug":92},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Exam",70,"exam",{"id":98,"doc_module":4,"doc_module_name":46,"category_name":99,"show_sort_weight":100,"slug":101},5,"Comic",60,"comic",{"id":103,"doc_module":4,"doc_module_name":46,"category_name":104,"show_sort_weight":105,"slug":106},6,"Technology",50,"technology",{"id":108,"doc_module":4,"doc_module_name":46,"category_name":109,"show_sort_weight":110,"slug":111},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":113,"slug":114},30,"research-report",{"id":116,"doc_module":4,"doc_module_name":46,"category_name":117,"show_sort_weight":118,"slug":119},9,"Religion & Spirituality",20,"religion-spirituality",{"id":118,"doc_module":4,"doc_module_name":46,"category_name":121,"show_sort_weight":118,"slug":122},"World Cup","world-cup",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":125,"show_sort_weight":124,"slug":126},10,"Lifestyle","lifestyle",{"id":128,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":98,"slug":130},19,"General","general"]