Lookahead Counterfactual Fairness

As machine learning (ML) algorithms are used in applications that involve humans, concerns have arisen that these algorithms may be biased against certain social groups. \textit{Counterfactual fairness} (CF) is a fairness notion proposed in Kusner et al. (2017) that measures the unfairness of ML pre...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Zuo, Zhiqun, Xie, Tian, Tan, Xuwei, Zhang, Xueru, Khalili, Mohammad Mahdi
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title
container_volume
creator Zuo, Zhiqun
Xie, Tian
Tan, Xuwei
Zhang, Xueru
Khalili, Mohammad Mahdi
description As machine learning (ML) algorithms are used in applications that involve humans, concerns have arisen that these algorithms may be biased against certain social groups. \textit{Counterfactual fairness} (CF) is a fairness notion proposed in Kusner et al. (2017) that measures the unfairness of ML predictions; it requires that the prediction perceived by an individual in the real world has the same marginal distribution as it would be in a counterfactual world, in which the individual belongs to a different group. Although CF ensures fair ML predictions, it fails to consider the downstream effects of ML predictions on individuals. Since humans are strategic and often adapt their behaviors in response to the ML system, predictions that satisfy CF may not lead to a fair future outcome for the individuals. In this paper, we introduce \textit{lookahead counterfactual fairness} (LCF), a fairness notion accounting for the downstream effects of ML models which requires the individual \textit{future status} to be counterfactually fair. We theoretically identify conditions under which LCF can be satisfied and propose an algorithm based on the theorems. We also extend the concept to path-dependent fairness. Experiments on both synthetic and real data validate the proposed method.
doi_str_mv 10.48550/arxiv.2412.01065
format Article
fullrecord <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_2412_01065</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2412_01065</sourcerecordid><originalsourceid>FETCH-arxiv_primary_2412_010653</originalsourceid><addsrcrecordid>eNpjYJA0NNAzsTA1NdBPLKrILNMzMjE00jMwNDAz5WRQ9MnPz07MSE1MUXDOL80rSS1KS0wuKU3MUXBLzCzKSy0u5mFgTUvMKU7lhdLcDPJuriHOHrpgw-ILijJzE4sq40GGxoMNNSasAgA_dytX</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Lookahead Counterfactual Fairness</title><source>arXiv.org</source><creator>Zuo, Zhiqun ; Xie, Tian ; Tan, Xuwei ; Zhang, Xueru ; Khalili, Mohammad Mahdi</creator><creatorcontrib>Zuo, Zhiqun ; Xie, Tian ; Tan, Xuwei ; Zhang, Xueru ; Khalili, Mohammad Mahdi</creatorcontrib><description>As machine learning (ML) algorithms are used in applications that involve humans, concerns have arisen that these algorithms may be biased against certain social groups. \textit{Counterfactual fairness} (CF) is a fairness notion proposed in Kusner et al. (2017) that measures the unfairness of ML predictions; it requires that the prediction perceived by an individual in the real world has the same marginal distribution as it would be in a counterfactual world, in which the individual belongs to a different group. Although CF ensures fair ML predictions, it fails to consider the downstream effects of ML predictions on individuals. Since humans are strategic and often adapt their behaviors in response to the ML system, predictions that satisfy CF may not lead to a fair future outcome for the individuals. In this paper, we introduce \textit{lookahead counterfactual fairness} (LCF), a fairness notion accounting for the downstream effects of ML models which requires the individual \textit{future status} to be counterfactually fair. We theoretically identify conditions under which LCF can be satisfied and propose an algorithm based on the theorems. We also extend the concept to path-dependent fairness. Experiments on both synthetic and real data validate the proposed method.</description><identifier>DOI: 10.48550/arxiv.2412.01065</identifier><language>eng</language><subject>Computer Science - Artificial Intelligence ; Computer Science - Learning ; Statistics - Machine Learning</subject><creationdate>2024-12</creationdate><rights>http://arxiv.org/licenses/nonexclusive-distrib/1.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,777,882</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/2412.01065$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.2412.01065$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Zuo, Zhiqun</creatorcontrib><creatorcontrib>Xie, Tian</creatorcontrib><creatorcontrib>Tan, Xuwei</creatorcontrib><creatorcontrib>Zhang, Xueru</creatorcontrib><creatorcontrib>Khalili, Mohammad Mahdi</creatorcontrib><title>Lookahead Counterfactual Fairness</title><description>As machine learning (ML) algorithms are used in applications that involve humans, concerns have arisen that these algorithms may be biased against certain social groups. \textit{Counterfactual fairness} (CF) is a fairness notion proposed in Kusner et al. (2017) that measures the unfairness of ML predictions; it requires that the prediction perceived by an individual in the real world has the same marginal distribution as it would be in a counterfactual world, in which the individual belongs to a different group. Although CF ensures fair ML predictions, it fails to consider the downstream effects of ML predictions on individuals. Since humans are strategic and often adapt their behaviors in response to the ML system, predictions that satisfy CF may not lead to a fair future outcome for the individuals. In this paper, we introduce \textit{lookahead counterfactual fairness} (LCF), a fairness notion accounting for the downstream effects of ML models which requires the individual \textit{future status} to be counterfactually fair. We theoretically identify conditions under which LCF can be satisfied and propose an algorithm based on the theorems. We also extend the concept to path-dependent fairness. Experiments on both synthetic and real data validate the proposed method.</description><subject>Computer Science - Artificial Intelligence</subject><subject>Computer Science - Learning</subject><subject>Statistics - Machine Learning</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2024</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNpjYJA0NNAzsTA1NdBPLKrILNMzMjE00jMwNDAz5WRQ9MnPz07MSE1MUXDOL80rSS1KS0wuKU3MUXBLzCzKSy0u5mFgTUvMKU7lhdLcDPJuriHOHrpgw-ILijJzE4sq40GGxoMNNSasAgA_dytX</recordid><startdate>20241201</startdate><enddate>20241201</enddate><creator>Zuo, Zhiqun</creator><creator>Xie, Tian</creator><creator>Tan, Xuwei</creator><creator>Zhang, Xueru</creator><creator>Khalili, Mohammad Mahdi</creator><scope>AKY</scope><scope>EPD</scope><scope>GOX</scope></search><sort><creationdate>20241201</creationdate><title>Lookahead Counterfactual Fairness</title><author>Zuo, Zhiqun ; Xie, Tian ; Tan, Xuwei ; Zhang, Xueru ; Khalili, Mohammad Mahdi</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-arxiv_primary_2412_010653</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2024</creationdate><topic>Computer Science - Artificial Intelligence</topic><topic>Computer Science - Learning</topic><topic>Statistics - Machine Learning</topic><toplevel>online_resources</toplevel><creatorcontrib>Zuo, Zhiqun</creatorcontrib><creatorcontrib>Xie, Tian</creatorcontrib><creatorcontrib>Tan, Xuwei</creatorcontrib><creatorcontrib>Zhang, Xueru</creatorcontrib><creatorcontrib>Khalili, Mohammad Mahdi</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv Statistics</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Zuo, Zhiqun</au><au>Xie, Tian</au><au>Tan, Xuwei</au><au>Zhang, Xueru</au><au>Khalili, Mohammad Mahdi</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Lookahead Counterfactual Fairness</atitle><date>2024-12-01</date><risdate>2024</risdate><abstract>As machine learning (ML) algorithms are used in applications that involve humans, concerns have arisen that these algorithms may be biased against certain social groups. \textit{Counterfactual fairness} (CF) is a fairness notion proposed in Kusner et al. (2017) that measures the unfairness of ML predictions; it requires that the prediction perceived by an individual in the real world has the same marginal distribution as it would be in a counterfactual world, in which the individual belongs to a different group. Although CF ensures fair ML predictions, it fails to consider the downstream effects of ML predictions on individuals. Since humans are strategic and often adapt their behaviors in response to the ML system, predictions that satisfy CF may not lead to a fair future outcome for the individuals. In this paper, we introduce \textit{lookahead counterfactual fairness} (LCF), a fairness notion accounting for the downstream effects of ML models which requires the individual \textit{future status} to be counterfactually fair. We theoretically identify conditions under which LCF can be satisfied and propose an algorithm based on the theorems. We also extend the concept to path-dependent fairness. Experiments on both synthetic and real data validate the proposed method.</abstract><doi>10.48550/arxiv.2412.01065</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext_linktorsrc
identifier DOI: 10.48550/arxiv.2412.01065
ispartof
issn
language eng
recordid cdi_arxiv_primary_2412_01065
source arXiv.org
subjects Computer Science - Artificial Intelligence
Computer Science - Learning
Statistics - Machine Learning
title Lookahead Counterfactual Fairness
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-17T22%3A24%3A12IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-arxiv_GOX&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Lookahead%20Counterfactual%20Fairness&rft.au=Zuo,%20Zhiqun&rft.date=2024-12-01&rft_id=info:doi/10.48550/arxiv.2412.01065&rft_dat=%3Carxiv_GOX%3E2412_01065%3C/arxiv_GOX%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true