MoCoUTRL: a momentum contrastive framework for unsupervised text representation learning

This paper presents MoCoUTRL: a Momentum Contrastive Framework for Unsupervised Text Representation Learning. This model improves two aspects of recently popular contrastive learning algorithms in natural language processing (NLP). Firstly, MoCoUTRL employs multi-granularity semantic contrastive lea...

Full description

Saved in:
Bibliographic Details
Main Authors: Ao Zou, Wenning Hao, Dawei Jin, Gang Chen, Feiyan Sun
Format: Article
Language:English
Published: Taylor & Francis Group 2023-12-01
Series:Connection Science
Subjects:
Online Access:http://dx.doi.org/10.1080/09540091.2023.2221406
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850211082404102144
author Ao Zou
Wenning Hao
Dawei Jin
Gang Chen
Feiyan Sun
author_facet Ao Zou
Wenning Hao
Dawei Jin
Gang Chen
Feiyan Sun
author_sort Ao Zou
collection DOAJ
description This paper presents MoCoUTRL: a Momentum Contrastive Framework for Unsupervised Text Representation Learning. This model improves two aspects of recently popular contrastive learning algorithms in natural language processing (NLP). Firstly, MoCoUTRL employs multi-granularity semantic contrastive learning objectives, enabling a more comprehensive understanding of the semantic features of samples. Secondly, MoCoUTRL uses a dynamic dictionary to act as the approximately ground-truth representation for each token, providing the pseudo labels for token-level contrastive learning. The MoCoUTRL can extend the use of pre-trained language models (PLM) and even large-scale language models (LLM) into a plug-and-play semantic feature extractor that can fuel multiple downstream tasks. Experimental results on several publicly available datasets and further theoretical analysis validate the effectiveness and interpretability of the proposed method in this paper.
format Article
id doaj-art-8d9dceb6c5ce4e9fa3fc2acb9be0756a
institution OA Journals
issn 0954-0091
1360-0494
language English
publishDate 2023-12-01
publisher Taylor & Francis Group
record_format Article
series Connection Science
spelling doaj-art-8d9dceb6c5ce4e9fa3fc2acb9be0756a2025-08-20T02:09:38ZengTaylor & Francis GroupConnection Science0954-00911360-04942023-12-0135110.1080/09540091.2023.22214062221406MoCoUTRL: a momentum contrastive framework for unsupervised text representation learningAo Zou0Wenning Hao1Dawei Jin2Gang Chen3Feiyan Sun4Command and Control Engineering College, Army Engineering University of PLACommand and Control Engineering College, Army Engineering University of PLACommand and Control Engineering College, Army Engineering University of PLACommand and Control Engineering College, Army Engineering University of PLACommand and Control Engineering College, Army Engineering University of PLAThis paper presents MoCoUTRL: a Momentum Contrastive Framework for Unsupervised Text Representation Learning. This model improves two aspects of recently popular contrastive learning algorithms in natural language processing (NLP). Firstly, MoCoUTRL employs multi-granularity semantic contrastive learning objectives, enabling a more comprehensive understanding of the semantic features of samples. Secondly, MoCoUTRL uses a dynamic dictionary to act as the approximately ground-truth representation for each token, providing the pseudo labels for token-level contrastive learning. The MoCoUTRL can extend the use of pre-trained language models (PLM) and even large-scale language models (LLM) into a plug-and-play semantic feature extractor that can fuel multiple downstream tasks. Experimental results on several publicly available datasets and further theoretical analysis validate the effectiveness and interpretability of the proposed method in this paper.http://dx.doi.org/10.1080/09540091.2023.2221406natural language processingtext representation learningmomentum contrastalignmentuniformity
spellingShingle Ao Zou
Wenning Hao
Dawei Jin
Gang Chen
Feiyan Sun
MoCoUTRL: a momentum contrastive framework for unsupervised text representation learning
Connection Science
natural language processing
text representation learning
momentum contrast
alignment
uniformity
title MoCoUTRL: a momentum contrastive framework for unsupervised text representation learning
title_full MoCoUTRL: a momentum contrastive framework for unsupervised text representation learning
title_fullStr MoCoUTRL: a momentum contrastive framework for unsupervised text representation learning
title_full_unstemmed MoCoUTRL: a momentum contrastive framework for unsupervised text representation learning
title_short MoCoUTRL: a momentum contrastive framework for unsupervised text representation learning
title_sort mocoutrl a momentum contrastive framework for unsupervised text representation learning
topic natural language processing
text representation learning
momentum contrast
alignment
uniformity
url http://dx.doi.org/10.1080/09540091.2023.2221406
work_keys_str_mv AT aozou mocoutrlamomentumcontrastiveframeworkforunsupervisedtextrepresentationlearning
AT wenninghao mocoutrlamomentumcontrastiveframeworkforunsupervisedtextrepresentationlearning
AT daweijin mocoutrlamomentumcontrastiveframeworkforunsupervisedtextrepresentationlearning
AT gangchen mocoutrlamomentumcontrastiveframeworkforunsupervisedtextrepresentationlearning
AT feiyansun mocoutrlamomentumcontrastiveframeworkforunsupervisedtextrepresentationlearning