@inproceedings{igbaria-belinkov-2024-learning,
title = "Learning from Others: Similarity-based Regularization for Mitigating Dataset Bias.",
author = "Igbaria, Reda and
Belinkov, Yonatan",
editor = "Zhao, Chen and
Mosbach, Marius and
Atanasova, Pepa and
Goldfarb-Tarrent, Seraphina and
Hase, Peter and
Hosseini, Arian and
Elbayad, Maha and
Pezzelle, Sandro and
Mozes, Maximilian",
booktitle = "Proceedings of the 9th Workshop on Representation Learning for NLP (RepL4NLP-2024)",
month = aug,
year = "2024",
address = "Bangkok, Thailand",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.repl4nlp-1.4",
pages = "37--50",
abstract = "Common methods for mitigating spurious correlations in natural language understanding (NLU) usually operate in the output space, encouraging a main model to behave differently from a bias model by down-weighing examples where the bias model is confident.While improving out of distribution (OOD) performance, it was recently observed that the internal representations of the presumably debiased models are actually more, rather than less biased. We propose SimgReg, a new method for debiasing internal model components via similarity-based regularization, in representation space: We encourage the model to learn representations that are either similar to an unbiased model or different from a biased model. We experiment with three NLU tasks and different kinds of biases.We find that SimReg improves OOD performance, with little in-distribution degradation. Moreover, the representations learned by SimReg are less biased than in other methods.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="igbaria-belinkov-2024-learning">
<titleInfo>
<title>Learning from Others: Similarity-based Regularization for Mitigating Dataset Bias.</title>
</titleInfo>
<name type="personal">
<namePart type="given">Reda</namePart>
<namePart type="family">Igbaria</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yonatan</namePart>
<namePart type="family">Belinkov</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2024-08</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 9th Workshop on Representation Learning for NLP (RepL4NLP-2024)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Chen</namePart>
<namePart type="family">Zhao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Marius</namePart>
<namePart type="family">Mosbach</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Pepa</namePart>
<namePart type="family">Atanasova</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Seraphina</namePart>
<namePart type="family">Goldfarb-Tarrent</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Peter</namePart>
<namePart type="family">Hase</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Arian</namePart>
<namePart type="family">Hosseini</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Maha</namePart>
<namePart type="family">Elbayad</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sandro</namePart>
<namePart type="family">Pezzelle</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Maximilian</namePart>
<namePart type="family">Mozes</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Bangkok, Thailand</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Common methods for mitigating spurious correlations in natural language understanding (NLU) usually operate in the output space, encouraging a main model to behave differently from a bias model by down-weighing examples where the bias model is confident.While improving out of distribution (OOD) performance, it was recently observed that the internal representations of the presumably debiased models are actually more, rather than less biased. We propose SimgReg, a new method for debiasing internal model components via similarity-based regularization, in representation space: We encourage the model to learn representations that are either similar to an unbiased model or different from a biased model. We experiment with three NLU tasks and different kinds of biases.We find that SimReg improves OOD performance, with little in-distribution degradation. Moreover, the representations learned by SimReg are less biased than in other methods.</abstract>
<identifier type="citekey">igbaria-belinkov-2024-learning</identifier>
<location>
<url>https://aclanthology.org/2024.repl4nlp-1.4</url>
</location>
<part>
<date>2024-08</date>
<extent unit="page">
<start>37</start>
<end>50</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Learning from Others: Similarity-based Regularization for Mitigating Dataset Bias.
%A Igbaria, Reda
%A Belinkov, Yonatan
%Y Zhao, Chen
%Y Mosbach, Marius
%Y Atanasova, Pepa
%Y Goldfarb-Tarrent, Seraphina
%Y Hase, Peter
%Y Hosseini, Arian
%Y Elbayad, Maha
%Y Pezzelle, Sandro
%Y Mozes, Maximilian
%S Proceedings of the 9th Workshop on Representation Learning for NLP (RepL4NLP-2024)
%D 2024
%8 August
%I Association for Computational Linguistics
%C Bangkok, Thailand
%F igbaria-belinkov-2024-learning
%X Common methods for mitigating spurious correlations in natural language understanding (NLU) usually operate in the output space, encouraging a main model to behave differently from a bias model by down-weighing examples where the bias model is confident.While improving out of distribution (OOD) performance, it was recently observed that the internal representations of the presumably debiased models are actually more, rather than less biased. We propose SimgReg, a new method for debiasing internal model components via similarity-based regularization, in representation space: We encourage the model to learn representations that are either similar to an unbiased model or different from a biased model. We experiment with three NLU tasks and different kinds of biases.We find that SimReg improves OOD performance, with little in-distribution degradation. Moreover, the representations learned by SimReg are less biased than in other methods.
%U https://aclanthology.org/2024.repl4nlp-1.4
%P 37-50
Markdown (Informal)
[Learning from Others: Similarity-based Regularization for Mitigating Dataset Bias.](https://aclanthology.org/2024.repl4nlp-1.4) (Igbaria & Belinkov, RepL4NLP-WS 2024)
ACL