@inproceedings{dong-etal-2022-cml,
title = "{CML}: A Contrastive Meta Learning Method to Estimate Human Label Confidence Scores and Reduce Data Collection Cost",
author = "Dong, Bo and
Wang, Yiyi and
Sun, Hanbo and
Wang, Yunji and
Hashemi, Alireza and
Du, Zheng",
editor = "Malmasi, Shervin and
Rokhlenko, Oleg and
Ueffing, Nicola and
Guy, Ido and
Agichtein, Eugene and
Kallumadi, Surya",
booktitle = "Proceedings of the Fifth Workshop on e-Commerce and NLP (ECNLP 5)",
month = may,
year = "2022",
address = "Dublin, Ireland",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2022.ecnlp-1.5",
doi = "10.18653/v1/2022.ecnlp-1.5",
pages = "35--43",
abstract = "Deep neural network models are especially susceptible to noise in annotated labels. In the real world, annotated data typically contains noise caused by a variety of factors such as task difficulty, annotator experience, and annotator bias. Label quality is critical for label validation tasks; however, correcting for noise by collecting more data is often costly. In this paper, we propose a contrastive meta-learning framework (CML) to address the challenges introduced by noisy annotated data, specifically in the context of natural language processing. CML combines contrastive and meta learning to improve the quality of text feature representations. Meta-learning is also used to generate confidence scores to assess label quality. We demonstrate that a model built on CML-filtered data outperforms a model built on clean data. Furthermore, we perform experiments on deidentified commercial voice assistant datasets and demonstrate that our model outperforms several SOTA approaches.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="dong-etal-2022-cml">
<titleInfo>
<title>CML: A Contrastive Meta Learning Method to Estimate Human Label Confidence Scores and Reduce Data Collection Cost</title>
</titleInfo>
<name type="personal">
<namePart type="given">Bo</namePart>
<namePart type="family">Dong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yiyi</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hanbo</namePart>
<namePart type="family">Sun</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yunji</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Alireza</namePart>
<namePart type="family">Hashemi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zheng</namePart>
<namePart type="family">Du</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2022-05</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the Fifth Workshop on e-Commerce and NLP (ECNLP 5)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Shervin</namePart>
<namePart type="family">Malmasi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Oleg</namePart>
<namePart type="family">Rokhlenko</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Nicola</namePart>
<namePart type="family">Ueffing</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ido</namePart>
<namePart type="family">Guy</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Eugene</namePart>
<namePart type="family">Agichtein</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Surya</namePart>
<namePart type="family">Kallumadi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Dublin, Ireland</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Deep neural network models are especially susceptible to noise in annotated labels. In the real world, annotated data typically contains noise caused by a variety of factors such as task difficulty, annotator experience, and annotator bias. Label quality is critical for label validation tasks; however, correcting for noise by collecting more data is often costly. In this paper, we propose a contrastive meta-learning framework (CML) to address the challenges introduced by noisy annotated data, specifically in the context of natural language processing. CML combines contrastive and meta learning to improve the quality of text feature representations. Meta-learning is also used to generate confidence scores to assess label quality. We demonstrate that a model built on CML-filtered data outperforms a model built on clean data. Furthermore, we perform experiments on deidentified commercial voice assistant datasets and demonstrate that our model outperforms several SOTA approaches.</abstract>
<identifier type="citekey">dong-etal-2022-cml</identifier>
<identifier type="doi">10.18653/v1/2022.ecnlp-1.5</identifier>
<location>
<url>https://aclanthology.org/2022.ecnlp-1.5</url>
</location>
<part>
<date>2022-05</date>
<extent unit="page">
<start>35</start>
<end>43</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T CML: A Contrastive Meta Learning Method to Estimate Human Label Confidence Scores and Reduce Data Collection Cost
%A Dong, Bo
%A Wang, Yiyi
%A Sun, Hanbo
%A Wang, Yunji
%A Hashemi, Alireza
%A Du, Zheng
%Y Malmasi, Shervin
%Y Rokhlenko, Oleg
%Y Ueffing, Nicola
%Y Guy, Ido
%Y Agichtein, Eugene
%Y Kallumadi, Surya
%S Proceedings of the Fifth Workshop on e-Commerce and NLP (ECNLP 5)
%D 2022
%8 May
%I Association for Computational Linguistics
%C Dublin, Ireland
%F dong-etal-2022-cml
%X Deep neural network models are especially susceptible to noise in annotated labels. In the real world, annotated data typically contains noise caused by a variety of factors such as task difficulty, annotator experience, and annotator bias. Label quality is critical for label validation tasks; however, correcting for noise by collecting more data is often costly. In this paper, we propose a contrastive meta-learning framework (CML) to address the challenges introduced by noisy annotated data, specifically in the context of natural language processing. CML combines contrastive and meta learning to improve the quality of text feature representations. Meta-learning is also used to generate confidence scores to assess label quality. We demonstrate that a model built on CML-filtered data outperforms a model built on clean data. Furthermore, we perform experiments on deidentified commercial voice assistant datasets and demonstrate that our model outperforms several SOTA approaches.
%R 10.18653/v1/2022.ecnlp-1.5
%U https://aclanthology.org/2022.ecnlp-1.5
%U https://doi.org/10.18653/v1/2022.ecnlp-1.5
%P 35-43
Markdown (Informal)
[CML: A Contrastive Meta Learning Method to Estimate Human Label Confidence Scores and Reduce Data Collection Cost](https://aclanthology.org/2022.ecnlp-1.5) (Dong et al., ECNLP 2022)
ACL