Detecting Underspecification with Local Ensembles

We present local ensembles, a method for detecting underspecification -- when many possible predictors are consistent with the training data and model class -- at test time in a pre-trained model. Our method uses local second-order information to approximate the variance of predictions across an ens...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:arXiv.org 2021-12
Hauptverfasser: Madras, David, Atwood, James, D'Amour, Alex
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title arXiv.org
container_volume
creator Madras, David
Atwood, James
D'Amour, Alex
description We present local ensembles, a method for detecting underspecification -- when many possible predictors are consistent with the training data and model class -- at test time in a pre-trained model. Our method uses local second-order information to approximate the variance of predictions across an ensemble of models from the same class. We compute this approximation by estimating the norm of the component of a test point's gradient that aligns with the low-curvature directions of the Hessian, and provide a tractable method for estimating this quantity. Experimentally, we show that our method is capable of detecting when a pre-trained model is underspecified on test data, with applications to out-of-distribution detection, detecting spurious correlates, and active learning.
format Article
fullrecord <record><control><sourceid>proquest</sourceid><recordid>TN_cdi_proquest_journals_2307882237</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2307882237</sourcerecordid><originalsourceid>FETCH-proquest_journals_23078822373</originalsourceid><addsrcrecordid>eNqNyrEKwjAUQNEgCBbtPwScC-mLNdm14uCoc4nxVVNiUvNS_H0d_ACnO5w7YwVIWVd6A7BgJdEghICtgqaRBav3mNFmF-78Em6YaETremdNdjHwt8sPforWeN4GwufVI63YvDeesPx1ydaH9rw7VmOKrwkpd0OcUvhSB1IorQGkkv9dH8p7M8U</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2307882237</pqid></control><display><type>article</type><title>Detecting Underspecification with Local Ensembles</title><source>Free E- Journals</source><creator>Madras, David ; Atwood, James ; D'Amour, Alex</creator><creatorcontrib>Madras, David ; Atwood, James ; D'Amour, Alex</creatorcontrib><description>We present local ensembles, a method for detecting underspecification -- when many possible predictors are consistent with the training data and model class -- at test time in a pre-trained model. Our method uses local second-order information to approximate the variance of predictions across an ensemble of models from the same class. We compute this approximation by estimating the norm of the component of a test point's gradient that aligns with the low-curvature directions of the Hessian, and provide a tractable method for estimating this quantity. Experimentally, we show that our method is capable of detecting when a pre-trained model is underspecified on test data, with applications to out-of-distribution detection, detecting spurious correlates, and active learning.</description><identifier>EISSN: 2331-8422</identifier><language>eng</language><publisher>Ithaca: Cornell University Library, arXiv.org</publisher><subject>Estimation ; Extrapolation</subject><ispartof>arXiv.org, 2021-12</ispartof><rights>2021. This work is published under http://arxiv.org/licenses/nonexclusive-distrib/1.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.</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>776,780</link.rule.ids></links><search><creatorcontrib>Madras, David</creatorcontrib><creatorcontrib>Atwood, James</creatorcontrib><creatorcontrib>D'Amour, Alex</creatorcontrib><title>Detecting Underspecification with Local Ensembles</title><title>arXiv.org</title><description>We present local ensembles, a method for detecting underspecification -- when many possible predictors are consistent with the training data and model class -- at test time in a pre-trained model. Our method uses local second-order information to approximate the variance of predictions across an ensemble of models from the same class. We compute this approximation by estimating the norm of the component of a test point's gradient that aligns with the low-curvature directions of the Hessian, and provide a tractable method for estimating this quantity. Experimentally, we show that our method is capable of detecting when a pre-trained model is underspecified on test data, with applications to out-of-distribution detection, detecting spurious correlates, and active learning.</description><subject>Estimation</subject><subject>Extrapolation</subject><issn>2331-8422</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2021</creationdate><recordtype>article</recordtype><sourceid>BENPR</sourceid><recordid>eNqNyrEKwjAUQNEgCBbtPwScC-mLNdm14uCoc4nxVVNiUvNS_H0d_ACnO5w7YwVIWVd6A7BgJdEghICtgqaRBav3mNFmF-78Em6YaETremdNdjHwt8sPforWeN4GwufVI63YvDeesPx1ydaH9rw7VmOKrwkpd0OcUvhSB1IorQGkkv9dH8p7M8U</recordid><startdate>20211207</startdate><enddate>20211207</enddate><creator>Madras, David</creator><creator>Atwood, James</creator><creator>D'Amour, Alex</creator><general>Cornell University Library, arXiv.org</general><scope>8FE</scope><scope>8FG</scope><scope>ABJCF</scope><scope>ABUWG</scope><scope>AFKRA</scope><scope>AZQEC</scope><scope>BENPR</scope><scope>BGLVJ</scope><scope>CCPQU</scope><scope>DWQXO</scope><scope>HCIFZ</scope><scope>L6V</scope><scope>M7S</scope><scope>PIMPY</scope><scope>PQEST</scope><scope>PQQKQ</scope><scope>PQUKI</scope><scope>PRINS</scope><scope>PTHSS</scope></search><sort><creationdate>20211207</creationdate><title>Detecting Underspecification with Local Ensembles</title><author>Madras, David ; Atwood, James ; D'Amour, Alex</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-proquest_journals_23078822373</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2021</creationdate><topic>Estimation</topic><topic>Extrapolation</topic><toplevel>online_resources</toplevel><creatorcontrib>Madras, David</creatorcontrib><creatorcontrib>Atwood, James</creatorcontrib><creatorcontrib>D'Amour, Alex</creatorcontrib><collection>ProQuest SciTech Collection</collection><collection>ProQuest Technology Collection</collection><collection>Materials Science &amp; Engineering Collection</collection><collection>ProQuest Central (Alumni Edition)</collection><collection>ProQuest Central UK/Ireland</collection><collection>ProQuest Central Essentials</collection><collection>ProQuest Central</collection><collection>Technology Collection</collection><collection>ProQuest One Community College</collection><collection>ProQuest Central Korea</collection><collection>SciTech Premium Collection</collection><collection>ProQuest Engineering Collection</collection><collection>Engineering Database</collection><collection>Publicly Available Content Database</collection><collection>ProQuest One Academic Eastern Edition (DO NOT USE)</collection><collection>ProQuest One Academic</collection><collection>ProQuest One Academic UKI Edition</collection><collection>ProQuest Central China</collection><collection>Engineering Collection</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Madras, David</au><au>Atwood, James</au><au>D'Amour, Alex</au><format>book</format><genre>document</genre><ristype>GEN</ristype><atitle>Detecting Underspecification with Local Ensembles</atitle><jtitle>arXiv.org</jtitle><date>2021-12-07</date><risdate>2021</risdate><eissn>2331-8422</eissn><abstract>We present local ensembles, a method for detecting underspecification -- when many possible predictors are consistent with the training data and model class -- at test time in a pre-trained model. Our method uses local second-order information to approximate the variance of predictions across an ensemble of models from the same class. We compute this approximation by estimating the norm of the component of a test point's gradient that aligns with the low-curvature directions of the Hessian, and provide a tractable method for estimating this quantity. Experimentally, we show that our method is capable of detecting when a pre-trained model is underspecified on test data, with applications to out-of-distribution detection, detecting spurious correlates, and active learning.</abstract><cop>Ithaca</cop><pub>Cornell University Library, arXiv.org</pub><oa>free_for_read</oa></addata></record>
fulltext fulltext
identifier EISSN: 2331-8422
ispartof arXiv.org, 2021-12
issn 2331-8422
language eng
recordid cdi_proquest_journals_2307882237
source Free E- Journals
subjects Estimation
Extrapolation
title Detecting Underspecification with Local Ensembles
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-02-10T14%3A17%3A28IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest&rft_val_fmt=info:ofi/fmt:kev:mtx:book&rft.genre=document&rft.atitle=Detecting%20Underspecification%20with%20Local%20Ensembles&rft.jtitle=arXiv.org&rft.au=Madras,%20David&rft.date=2021-12-07&rft.eissn=2331-8422&rft_id=info:doi/&rft_dat=%3Cproquest%3E2307882237%3C/proquest%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=2307882237&rft_id=info:pmid/&rfr_iscdi=true