Conditionally Risk-Averse Contextual Bandits

Conditionally Risk-Averse Contextual Bandits

Contextual bandits with average-case statistical guarantees are inadequate in risk-averse situations because they might trade off degraded worst-case behaviour for better average performance. Designing a risk-averse contextual bandit is challenging because exploration is necessary but risk-aversion...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Farsang, Mónika, Mineiro, Paul, Zhang, Wangda
Format: Artikel
Sprache:eng
Schlagworte:
Computer Science - Learning
Statistics - Machine Learning
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 Farsang, Mónika
Mineiro, Paul
Zhang, Wangda
description Contextual bandits with average-case statistical guarantees are inadequate in risk-averse situations because they might trade off degraded worst-case behaviour for better average performance. Designing a risk-averse contextual bandit is challenging because exploration is necessary but risk-aversion is sensitive to the entire distribution of rewards; nonetheless we exhibit the first risk-averse contextual bandit algorithm with an online regret guarantee. We conduct experiments from diverse scenarios where worst-case outcomes should be avoided, from dynamic pricing, inventory management, and self-tuning software; including a production exascale data processing system.
doi_str_mv 10.48550/arxiv.2210.13573
format Article
fullrecord <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_2210_13573</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2210_13573</sourcerecordid><originalsourceid>FETCH-LOGICAL-a673-63fe5e9ef94f2554019d1979339d83662d005c547893fd8149eeac5988cfd7183</originalsourceid><addsrcrecordid>eNotjsFqwzAQRHXpoaT9gJ7qD4hdyau1tEfHNGkhECi5G2GtQMSNi-2Y5O9rpz0NzDyGJ8SLkpm2iPLN9dc4ZXk-FwrQwKNYV93ZxzF2Z9e2t-QrDqe0nLgfOJmXka_jxbXJxi3Q8CQegmsHfv7PlThu34_VR7o_7D6rcp-6wkBaQGBk4kA65IhaKvKKDAGQt1AUuZcSG9TGEgRvlSZm1yBZ2wRvlIWVeP27vevWP338dv2tXrTruzb8AuglO8M</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Conditionally Risk-Averse Contextual Bandits</title><source>arXiv.org</source><creator>Farsang, Mónika ; Mineiro, Paul ; Zhang, Wangda</creator><creatorcontrib>Farsang, Mónika ; Mineiro, Paul ; Zhang, Wangda</creatorcontrib><description>Contextual bandits with average-case statistical guarantees are inadequate in risk-averse situations because they might trade off degraded worst-case behaviour for better average performance. Designing a risk-averse contextual bandit is challenging because exploration is necessary but risk-aversion is sensitive to the entire distribution of rewards; nonetheless we exhibit the first risk-averse contextual bandit algorithm with an online regret guarantee. We conduct experiments from diverse scenarios where worst-case outcomes should be avoided, from dynamic pricing, inventory management, and self-tuning software; including a production exascale data processing system.</description><identifier>DOI: 10.48550/arxiv.2210.13573</identifier><language>eng</language><subject>Computer Science - Learning ; Statistics - Machine Learning</subject><creationdate>2022-10</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>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,776,881</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/2210.13573$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.2210.13573$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Farsang, Mónika</creatorcontrib><creatorcontrib>Mineiro, Paul</creatorcontrib><creatorcontrib>Zhang, Wangda</creatorcontrib><title>Conditionally Risk-Averse Contextual Bandits</title><description>Contextual bandits with average-case statistical guarantees are inadequate in risk-averse situations because they might trade off degraded worst-case behaviour for better average performance. Designing a risk-averse contextual bandit is challenging because exploration is necessary but risk-aversion is sensitive to the entire distribution of rewards; nonetheless we exhibit the first risk-averse contextual bandit algorithm with an online regret guarantee. We conduct experiments from diverse scenarios where worst-case outcomes should be avoided, from dynamic pricing, inventory management, and self-tuning software; including a production exascale data processing system.</description><subject>Computer Science - Learning</subject><subject>Statistics - Machine Learning</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2022</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNotjsFqwzAQRHXpoaT9gJ7qD4hdyau1tEfHNGkhECi5G2GtQMSNi-2Y5O9rpz0NzDyGJ8SLkpm2iPLN9dc4ZXk-FwrQwKNYV93ZxzF2Z9e2t-QrDqe0nLgfOJmXka_jxbXJxi3Q8CQegmsHfv7PlThu34_VR7o_7D6rcp-6wkBaQGBk4kA65IhaKvKKDAGQt1AUuZcSG9TGEgRvlSZm1yBZ2wRvlIWVeP27vevWP338dv2tXrTruzb8AuglO8M</recordid><startdate>20221024</startdate><enddate>20221024</enddate><creator>Farsang, Mónika</creator><creator>Mineiro, Paul</creator><creator>Zhang, Wangda</creator><scope>AKY</scope><scope>EPD</scope><scope>GOX</scope></search><sort><creationdate>20221024</creationdate><title>Conditionally Risk-Averse Contextual Bandits</title><author>Farsang, Mónika ; Mineiro, Paul ; Zhang, Wangda</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a673-63fe5e9ef94f2554019d1979339d83662d005c547893fd8149eeac5988cfd7183</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2022</creationdate><topic>Computer Science - Learning</topic><topic>Statistics - Machine Learning</topic><toplevel>online_resources</toplevel><creatorcontrib>Farsang, Mónika</creatorcontrib><creatorcontrib>Mineiro, Paul</creatorcontrib><creatorcontrib>Zhang, Wangda</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>Farsang, Mónika</au><au>Mineiro, Paul</au><au>Zhang, Wangda</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Conditionally Risk-Averse Contextual Bandits</atitle><date>2022-10-24</date><risdate>2022</risdate><abstract>Contextual bandits with average-case statistical guarantees are inadequate in risk-averse situations because they might trade off degraded worst-case behaviour for better average performance. Designing a risk-averse contextual bandit is challenging because exploration is necessary but risk-aversion is sensitive to the entire distribution of rewards; nonetheless we exhibit the first risk-averse contextual bandit algorithm with an online regret guarantee. We conduct experiments from diverse scenarios where worst-case outcomes should be avoided, from dynamic pricing, inventory management, and self-tuning software; including a production exascale data processing system.</abstract><doi>10.48550/arxiv.2210.13573</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext_linktorsrc
identifier DOI: 10.48550/arxiv.2210.13573
ispartof
issn
language eng
recordid cdi_arxiv_primary_2210_13573
source arXiv.org
subjects Computer Science - Learning
Statistics - Machine Learning
title Conditionally Risk-Averse Contextual Bandits
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-29T11%3A27%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=Conditionally%20Risk-Averse%20Contextual%20Bandits&rft.au=Farsang,%20M%C3%B3nika&rft.date=2022-10-24&rft_id=info:doi/10.48550/arxiv.2210.13573&rft_dat=%3Carxiv_GOX%3E2210_13573%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