End-to-End Self-Debiasing Framework for Robust NLU Training

Existing Natural Language Understanding (NLU) models have been shown to incorporate dataset biases leading to strong performance on in-distribution (ID) test sets but poor performance on out-of-distribution (OOD) ones. We introduce a simple yet effective debiasing framework whereby the shallow repre...

Full description

Saved in:
Bibliographic Details
Published in:arXiv.org 2021-09
Main Authors: Ghaddar, Abbas, Langlais, Philippe, Rezagholizadeh, Mehdi, Rashid, Ahmad
Format: Article
Language:English
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
cited_by
cites
container_end_page
container_issue
container_start_page
container_title arXiv.org
container_volume
creator Ghaddar, Abbas
Langlais, Philippe
Rezagholizadeh, Mehdi
Rashid, Ahmad
description Existing Natural Language Understanding (NLU) models have been shown to incorporate dataset biases leading to strong performance on in-distribution (ID) test sets but poor performance on out-of-distribution (OOD) ones. We introduce a simple yet effective debiasing framework whereby the shallow representations of the main model are used to derive a bias model and both models are trained simultaneously. We demonstrate on three well studied NLU tasks that despite its simplicity, our method leads to competitive OOD results. It significantly outperforms other debiasing approaches on two tasks, while still delivering high in-distribution performance.
doi_str_mv 10.48550/arxiv.2109.02071
format article
fullrecord <record><control><sourceid>proquest</sourceid><recordid>TN_cdi_proquest_journals_2569853262</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2569853262</sourcerecordid><originalsourceid>FETCH-LOGICAL-a522-3ef1a899fe5981b85b3b87f87a64a25a473a0f17a84db22f42bf6e8bd51b42b43</originalsourceid><addsrcrecordid>eNotjc1KxDAURoMgOIzzAO4CrlOTm9_iSsYZRygKWtfDDU2k49ho0qqPb0FX51scvkPIheCVclrzK8w__VcFgtcVB27FCVmAlII5BXBGVqUcOOdgLGgtF-R6M3RsTGwGfQ7HyG6D77H0wyvdZnwP3ym_0ZgyfUp-KiN9aF5om7EfZuOcnEY8lrD655K020273rHm8e5-fdMw1ABMhijQ1XUMunbCO-2ldzY6i0YhaFRWIo_ColOdB4gKfDTB-U4LP28ll-Ty7_Yjp88plHF_SFMe5uIetKmdlmBA_gLmF0fQ</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2569853262</pqid></control><display><type>article</type><title>End-to-End Self-Debiasing Framework for Robust NLU Training</title><source>Publicly Available Content Database</source><creator>Ghaddar, Abbas ; Langlais, Philippe ; Rezagholizadeh, Mehdi ; Rashid, Ahmad</creator><creatorcontrib>Ghaddar, Abbas ; Langlais, Philippe ; Rezagholizadeh, Mehdi ; Rashid, Ahmad</creatorcontrib><description>Existing Natural Language Understanding (NLU) models have been shown to incorporate dataset biases leading to strong performance on in-distribution (ID) test sets but poor performance on out-of-distribution (OOD) ones. We introduce a simple yet effective debiasing framework whereby the shallow representations of the main model are used to derive a bias model and both models are trained simultaneously. We demonstrate on three well studied NLU tasks that despite its simplicity, our method leads to competitive OOD results. It significantly outperforms other debiasing approaches on two tasks, while still delivering high in-distribution performance.</description><identifier>EISSN: 2331-8422</identifier><identifier>DOI: 10.48550/arxiv.2109.02071</identifier><language>eng</language><publisher>Ithaca: Cornell University Library, arXiv.org</publisher><ispartof>arXiv.org, 2021-09</ispartof><rights>2021. This work is published under http://creativecommons.org/licenses/by/4.0/ (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>$$Tsyndetics_thumb_exl</thumbnail><linktohtml>$$Uhttps://www.proquest.com/docview/2569853262?pq-origsite=primo$$EHTML$$P50$$Gproquest$$Hfree_for_read</linktohtml><link.rule.ids>780,784,25753,27925,37012,44590</link.rule.ids></links><search><creatorcontrib>Ghaddar, Abbas</creatorcontrib><creatorcontrib>Langlais, Philippe</creatorcontrib><creatorcontrib>Rezagholizadeh, Mehdi</creatorcontrib><creatorcontrib>Rashid, Ahmad</creatorcontrib><title>End-to-End Self-Debiasing Framework for Robust NLU Training</title><title>arXiv.org</title><description>Existing Natural Language Understanding (NLU) models have been shown to incorporate dataset biases leading to strong performance on in-distribution (ID) test sets but poor performance on out-of-distribution (OOD) ones. We introduce a simple yet effective debiasing framework whereby the shallow representations of the main model are used to derive a bias model and both models are trained simultaneously. We demonstrate on three well studied NLU tasks that despite its simplicity, our method leads to competitive OOD results. It significantly outperforms other debiasing approaches on two tasks, while still delivering high in-distribution performance.</description><issn>2331-8422</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2021</creationdate><recordtype>article</recordtype><sourceid>PIMPY</sourceid><recordid>eNotjc1KxDAURoMgOIzzAO4CrlOTm9_iSsYZRygKWtfDDU2k49ho0qqPb0FX51scvkPIheCVclrzK8w__VcFgtcVB27FCVmAlII5BXBGVqUcOOdgLGgtF-R6M3RsTGwGfQ7HyG6D77H0wyvdZnwP3ym_0ZgyfUp-KiN9aF5om7EfZuOcnEY8lrD655K020273rHm8e5-fdMw1ABMhijQ1XUMunbCO-2ldzY6i0YhaFRWIo_ColOdB4gKfDTB-U4LP28ll-Ty7_Yjp88plHF_SFMe5uIetKmdlmBA_gLmF0fQ</recordid><startdate>20210905</startdate><enddate>20210905</enddate><creator>Ghaddar, Abbas</creator><creator>Langlais, Philippe</creator><creator>Rezagholizadeh, Mehdi</creator><creator>Rashid, Ahmad</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>20210905</creationdate><title>End-to-End Self-Debiasing Framework for Robust NLU Training</title><author>Ghaddar, Abbas ; Langlais, Philippe ; Rezagholizadeh, Mehdi ; Rashid, Ahmad</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a522-3ef1a899fe5981b85b3b87f87a64a25a473a0f17a84db22f42bf6e8bd51b42b43</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2021</creationdate><toplevel>online_resources</toplevel><creatorcontrib>Ghaddar, Abbas</creatorcontrib><creatorcontrib>Langlais, Philippe</creatorcontrib><creatorcontrib>Rezagholizadeh, Mehdi</creatorcontrib><creatorcontrib>Rashid, Ahmad</creatorcontrib><collection>ProQuest SciTech Collection</collection><collection>ProQuest Technology Collection</collection><collection>Materials Science &amp; Engineering Collection</collection><collection>ProQuest Central (Alumni)</collection><collection>ProQuest Central</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><jtitle>arXiv.org</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Ghaddar, Abbas</au><au>Langlais, Philippe</au><au>Rezagholizadeh, Mehdi</au><au>Rashid, Ahmad</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>End-to-End Self-Debiasing Framework for Robust NLU Training</atitle><jtitle>arXiv.org</jtitle><date>2021-09-05</date><risdate>2021</risdate><eissn>2331-8422</eissn><abstract>Existing Natural Language Understanding (NLU) models have been shown to incorporate dataset biases leading to strong performance on in-distribution (ID) test sets but poor performance on out-of-distribution (OOD) ones. We introduce a simple yet effective debiasing framework whereby the shallow representations of the main model are used to derive a bias model and both models are trained simultaneously. We demonstrate on three well studied NLU tasks that despite its simplicity, our method leads to competitive OOD results. It significantly outperforms other debiasing approaches on two tasks, while still delivering high in-distribution performance.</abstract><cop>Ithaca</cop><pub>Cornell University Library, arXiv.org</pub><doi>10.48550/arxiv.2109.02071</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext
identifier EISSN: 2331-8422
ispartof arXiv.org, 2021-09
issn 2331-8422
language eng
recordid cdi_proquest_journals_2569853262
source Publicly Available Content Database
title End-to-End Self-Debiasing Framework for Robust NLU Training
url http://sfxeu10.hosted.exlibrisgroup.com/loughborough?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-29T15%3A08%3A43IST&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:journal&rft.genre=article&rft.atitle=End-to-End%20Self-Debiasing%20Framework%20for%20Robust%20NLU%20Training&rft.jtitle=arXiv.org&rft.au=Ghaddar,%20Abbas&rft.date=2021-09-05&rft.eissn=2331-8422&rft_id=info:doi/10.48550/arxiv.2109.02071&rft_dat=%3Cproquest%3E2569853262%3C/proquest%3E%3Cgrp_id%3Ecdi_FETCH-LOGICAL-a522-3ef1a899fe5981b85b3b87f87a64a25a473a0f17a84db22f42bf6e8bd51b42b43%3C/grp_id%3E%3Coa%3E%3C/oa%3E%3Curl%3E%3C/url%3E&rft_id=info:oai/&rft_pqid=2569853262&rft_id=info:pmid/&rfr_iscdi=true