@inproceedings{lu-etal-2021-attention,
title = "Attention Calibration for Transformer in Neural Machine Translation",
author = "Lu, Yu and
Zeng, Jiali and
Zhang, Jiajun and
Wu, Shuangzhi and
Li, Mu",
editor = "Zong, Chengqing and
Xia, Fei and
Li, Wenjie and
Navigli, Roberto",
booktitle = "Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers)",
month = aug,
year = "2021",
address = "Online",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2021.acl-long.103",
doi = "10.18653/v1/2021.acl-long.103",
pages = "1288--1298",
abstract = "Attention mechanisms have achieved substantial improvements in neural machine translation by dynamically selecting relevant inputs for different predictions. However, recent studies have questioned the attention mechanisms{'} capability for discovering decisive inputs. In this paper, we propose to calibrate the attention weights by introducing a mask perturbation model that automatically evaluates each input{'}s contribution to the model outputs. We increase the attention weights assigned to the indispensable tokens, whose removal leads to a dramatic performance decrease. The extensive experiments on the Transformer-based translation have demonstrated the effectiveness of our model. We further find that the calibrated attention weights are more uniform at lower layers to collect multiple information while more concentrated on the specific inputs at higher layers. Detailed analyses also show a great need for calibration in the attention weights with high entropy where the model is unconfident about its decision.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="lu-etal-2021-attention">
<titleInfo>
<title>Attention Calibration for Transformer in Neural Machine Translation</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yu</namePart>
<namePart type="family">Lu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiali</namePart>
<namePart type="family">Zeng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiajun</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shuangzhi</namePart>
<namePart type="family">Wu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mu</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2021-08</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Chengqing</namePart>
<namePart type="family">Zong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Fei</namePart>
<namePart type="family">Xia</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Wenjie</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Roberto</namePart>
<namePart type="family">Navigli</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Online</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Attention mechanisms have achieved substantial improvements in neural machine translation by dynamically selecting relevant inputs for different predictions. However, recent studies have questioned the attention mechanisms’ capability for discovering decisive inputs. In this paper, we propose to calibrate the attention weights by introducing a mask perturbation model that automatically evaluates each input’s contribution to the model outputs. We increase the attention weights assigned to the indispensable tokens, whose removal leads to a dramatic performance decrease. The extensive experiments on the Transformer-based translation have demonstrated the effectiveness of our model. We further find that the calibrated attention weights are more uniform at lower layers to collect multiple information while more concentrated on the specific inputs at higher layers. Detailed analyses also show a great need for calibration in the attention weights with high entropy where the model is unconfident about its decision.</abstract>
<identifier type="citekey">lu-etal-2021-attention</identifier>
<identifier type="doi">10.18653/v1/2021.acl-long.103</identifier>
<location>
<url>https://aclanthology.org/2021.acl-long.103</url>
</location>
<part>
<date>2021-08</date>
<extent unit="page">
<start>1288</start>
<end>1298</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Attention Calibration for Transformer in Neural Machine Translation
%A Lu, Yu
%A Zeng, Jiali
%A Zhang, Jiajun
%A Wu, Shuangzhi
%A Li, Mu
%Y Zong, Chengqing
%Y Xia, Fei
%Y Li, Wenjie
%Y Navigli, Roberto
%S Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers)
%D 2021
%8 August
%I Association for Computational Linguistics
%C Online
%F lu-etal-2021-attention
%X Attention mechanisms have achieved substantial improvements in neural machine translation by dynamically selecting relevant inputs for different predictions. However, recent studies have questioned the attention mechanisms’ capability for discovering decisive inputs. In this paper, we propose to calibrate the attention weights by introducing a mask perturbation model that automatically evaluates each input’s contribution to the model outputs. We increase the attention weights assigned to the indispensable tokens, whose removal leads to a dramatic performance decrease. The extensive experiments on the Transformer-based translation have demonstrated the effectiveness of our model. We further find that the calibrated attention weights are more uniform at lower layers to collect multiple information while more concentrated on the specific inputs at higher layers. Detailed analyses also show a great need for calibration in the attention weights with high entropy where the model is unconfident about its decision.
%R 10.18653/v1/2021.acl-long.103
%U https://aclanthology.org/2021.acl-long.103
%U https://doi.org/10.18653/v1/2021.acl-long.103
%P 1288-1298
Markdown (Informal)
[Attention Calibration for Transformer in Neural Machine Translation](https://aclanthology.org/2021.acl-long.103) (Lu et al., ACL-IJCNLP 2021)
ACL
- Yu Lu, Jiali Zeng, Jiajun Zhang, Shuangzhi Wu, and Mu Li. 2021. Attention Calibration for Transformer in Neural Machine Translation. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers), pages 1288–1298, Online. Association for Computational Linguistics.