Online Learning Using Only Peer Prediction

This paper considers a variant of the classical online learning problem with expert predictions. Our model's differences and challenges are due to lacking any direct feedback on the loss each expert incurs at each time step $t$. We propose an approach that uses peer prediction and identify cond...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Liu, Yang, Helmbold, David P
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 Liu, Yang
Helmbold, David P
description This paper considers a variant of the classical online learning problem with expert predictions. Our model's differences and challenges are due to lacking any direct feedback on the loss each expert incurs at each time step $t$. We propose an approach that uses peer prediction and identify conditions where it succeeds. Our techniques revolve around a carefully designed peer score function $s()$ that scores experts' predictions based on the peer consensus. We show a sufficient condition, that we call \emph{peer calibration}, under which standard online learning algorithms using loss feedback computed by the carefully crafted $s()$ have bounded regret with respect to the unrevealed ground truth values. We then demonstrate how suitable $s()$ functions can be derived for different assumptions and models.
doi_str_mv 10.48550/arxiv.1910.04382
format Article
fullrecord <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_1910_04382</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>1910_04382</sourcerecordid><originalsourceid>FETCH-LOGICAL-a672-e3a094718519ec57c3ad2a79449d82472540e69ac29eeae2d3632b01d8d444c73</originalsourceid><addsrcrecordid>eNotjruqAjEURdNYiPoBVk4tjOZxMklKkesDBsZC6-GYnHsJaJQoF_17n83esIrFYmwo-ASs1nyK-Rb_J8I9AQdlZZeNm3SIiYqaMKeY_ord5bVPei82RLnYZArRX-Mp9VnnFw8XGny_x7aLn-18VdbNcj2f1SVWRpakkDswwmrhyGvjFQaJxgG4YCUYqYFT5dBLR4Qkg6qU3HMRbAAAb1SPjT7ad2x7zvGI-d6-ott3tHoA_aM6mw</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Online Learning Using Only Peer Prediction</title><source>arXiv.org</source><creator>Liu, Yang ; Helmbold, David P</creator><creatorcontrib>Liu, Yang ; Helmbold, David P</creatorcontrib><description>This paper considers a variant of the classical online learning problem with expert predictions. Our model's differences and challenges are due to lacking any direct feedback on the loss each expert incurs at each time step $t$. We propose an approach that uses peer prediction and identify conditions where it succeeds. Our techniques revolve around a carefully designed peer score function $s()$ that scores experts' predictions based on the peer consensus. We show a sufficient condition, that we call \emph{peer calibration}, under which standard online learning algorithms using loss feedback computed by the carefully crafted $s()$ have bounded regret with respect to the unrevealed ground truth values. We then demonstrate how suitable $s()$ functions can be derived for different assumptions and models.</description><identifier>DOI: 10.48550/arxiv.1910.04382</identifier><language>eng</language><subject>Computer Science - Learning ; Statistics - Machine Learning</subject><creationdate>2019-10</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/1910.04382$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.1910.04382$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Liu, Yang</creatorcontrib><creatorcontrib>Helmbold, David P</creatorcontrib><title>Online Learning Using Only Peer Prediction</title><description>This paper considers a variant of the classical online learning problem with expert predictions. Our model's differences and challenges are due to lacking any direct feedback on the loss each expert incurs at each time step $t$. We propose an approach that uses peer prediction and identify conditions where it succeeds. Our techniques revolve around a carefully designed peer score function $s()$ that scores experts' predictions based on the peer consensus. We show a sufficient condition, that we call \emph{peer calibration}, under which standard online learning algorithms using loss feedback computed by the carefully crafted $s()$ have bounded regret with respect to the unrevealed ground truth values. We then demonstrate how suitable $s()$ functions can be derived for different assumptions and models.</description><subject>Computer Science - Learning</subject><subject>Statistics - Machine Learning</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2019</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNotjruqAjEURdNYiPoBVk4tjOZxMklKkesDBsZC6-GYnHsJaJQoF_17n83esIrFYmwo-ASs1nyK-Rb_J8I9AQdlZZeNm3SIiYqaMKeY_ord5bVPei82RLnYZArRX-Mp9VnnFw8XGny_x7aLn-18VdbNcj2f1SVWRpakkDswwmrhyGvjFQaJxgG4YCUYqYFT5dBLR4Qkg6qU3HMRbAAAb1SPjT7ad2x7zvGI-d6-ott3tHoA_aM6mw</recordid><startdate>20191010</startdate><enddate>20191010</enddate><creator>Liu, Yang</creator><creator>Helmbold, David P</creator><scope>AKY</scope><scope>EPD</scope><scope>GOX</scope></search><sort><creationdate>20191010</creationdate><title>Online Learning Using Only Peer Prediction</title><author>Liu, Yang ; Helmbold, David P</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a672-e3a094718519ec57c3ad2a79449d82472540e69ac29eeae2d3632b01d8d444c73</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2019</creationdate><topic>Computer Science - Learning</topic><topic>Statistics - Machine Learning</topic><toplevel>online_resources</toplevel><creatorcontrib>Liu, Yang</creatorcontrib><creatorcontrib>Helmbold, David P</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>Liu, Yang</au><au>Helmbold, David P</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Online Learning Using Only Peer Prediction</atitle><date>2019-10-10</date><risdate>2019</risdate><abstract>This paper considers a variant of the classical online learning problem with expert predictions. Our model's differences and challenges are due to lacking any direct feedback on the loss each expert incurs at each time step $t$. We propose an approach that uses peer prediction and identify conditions where it succeeds. Our techniques revolve around a carefully designed peer score function $s()$ that scores experts' predictions based on the peer consensus. We show a sufficient condition, that we call \emph{peer calibration}, under which standard online learning algorithms using loss feedback computed by the carefully crafted $s()$ have bounded regret with respect to the unrevealed ground truth values. We then demonstrate how suitable $s()$ functions can be derived for different assumptions and models.</abstract><doi>10.48550/arxiv.1910.04382</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext_linktorsrc
identifier DOI: 10.48550/arxiv.1910.04382
ispartof
issn
language eng
recordid cdi_arxiv_primary_1910_04382
source arXiv.org
subjects Computer Science - Learning
Statistics - Machine Learning
title Online Learning Using Only Peer Prediction
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-27T04%3A19%3A57IST&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=Online%20Learning%20Using%20Only%20Peer%20Prediction&rft.au=Liu,%20Yang&rft.date=2019-10-10&rft_id=info:doi/10.48550/arxiv.1910.04382&rft_dat=%3Carxiv_GOX%3E1910_04382%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