@inproceedings{chang-etal-2022-one,
title = "One-Teacher and Multiple-Student Knowledge Distillation on Sentiment Classification",
author = "Chang, Xiaoqin and
Lee, Sophia Yat Mei and
Zhu, Suyang and
Li, Shoushan and
Zhou, Guodong",
booktitle = "Proceedings of the 29th International Conference on Computational Linguistics",
month = oct,
year = "2022",
address = "Gyeongju, Republic of Korea",
publisher = "International Committee on Computational Linguistics",
url = "https://aclanthology.org/2022.coling-1.614",
pages = "7042--7052",
abstract = "Knowledge distillation is an effective method to transfer knowledge from a large pre-trained teacher model to a compacted student model. However, in previous studies, the distilled student models are still large and remain impractical in highly speed-sensitive systems (e.g., an IR system). In this study, we aim to distill a deep pre-trained model into an extremely compacted shallow model like CNN. Specifically, we propose a novel one-teacher and multiple-student knowledge distillation approach to distill a deep pre-trained teacher model into multiple shallow student models with ensemble learning. Moreover, we leverage large-scale unlabeled data to improve the performance of students. Empirical studies on three sentiment classification tasks demonstrate that our approach achieves better results with much fewer parameters (0.9{\%}-18{\%}) and extremely high speedup ratios (100X-1000X).",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="chang-etal-2022-one">
<titleInfo>
<title>One-Teacher and Multiple-Student Knowledge Distillation on Sentiment Classification</title>
</titleInfo>
<name type="personal">
<namePart type="given">Xiaoqin</namePart>
<namePart type="family">Chang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sophia</namePart>
<namePart type="given">Yat</namePart>
<namePart type="given">Mei</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Suyang</namePart>
<namePart type="family">Zhu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shoushan</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Guodong</namePart>
<namePart type="family">Zhou</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2022-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 29th International Conference on Computational Linguistics</title>
</titleInfo>
<originInfo>
<publisher>International Committee on Computational Linguistics</publisher>
<place>
<placeTerm type="text">Gyeongju, Republic of Korea</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Knowledge distillation is an effective method to transfer knowledge from a large pre-trained teacher model to a compacted student model. However, in previous studies, the distilled student models are still large and remain impractical in highly speed-sensitive systems (e.g., an IR system). In this study, we aim to distill a deep pre-trained model into an extremely compacted shallow model like CNN. Specifically, we propose a novel one-teacher and multiple-student knowledge distillation approach to distill a deep pre-trained teacher model into multiple shallow student models with ensemble learning. Moreover, we leverage large-scale unlabeled data to improve the performance of students. Empirical studies on three sentiment classification tasks demonstrate that our approach achieves better results with much fewer parameters (0.9%-18%) and extremely high speedup ratios (100X-1000X).</abstract>
<identifier type="citekey">chang-etal-2022-one</identifier>
<location>
<url>https://aclanthology.org/2022.coling-1.614</url>
</location>
<part>
<date>2022-10</date>
<extent unit="page">
<start>7042</start>
<end>7052</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T One-Teacher and Multiple-Student Knowledge Distillation on Sentiment Classification
%A Chang, Xiaoqin
%A Lee, Sophia Yat Mei
%A Zhu, Suyang
%A Li, Shoushan
%A Zhou, Guodong
%S Proceedings of the 29th International Conference on Computational Linguistics
%D 2022
%8 October
%I International Committee on Computational Linguistics
%C Gyeongju, Republic of Korea
%F chang-etal-2022-one
%X Knowledge distillation is an effective method to transfer knowledge from a large pre-trained teacher model to a compacted student model. However, in previous studies, the distilled student models are still large and remain impractical in highly speed-sensitive systems (e.g., an IR system). In this study, we aim to distill a deep pre-trained model into an extremely compacted shallow model like CNN. Specifically, we propose a novel one-teacher and multiple-student knowledge distillation approach to distill a deep pre-trained teacher model into multiple shallow student models with ensemble learning. Moreover, we leverage large-scale unlabeled data to improve the performance of students. Empirical studies on three sentiment classification tasks demonstrate that our approach achieves better results with much fewer parameters (0.9%-18%) and extremely high speedup ratios (100X-1000X).
%U https://aclanthology.org/2022.coling-1.614
%P 7042-7052
Markdown (Informal)
[One-Teacher and Multiple-Student Knowledge Distillation on Sentiment Classification](https://aclanthology.org/2022.coling-1.614) (Chang et al., COLING 2022)
ACL