Text Classification Model Enhanced by Unlabeled Data for LaTeX Formula

Generic language models pretrained on large unspecific domains are currently the foundation of NLP. Labeled data are limited in most model training due to the cost of manual annotation, especially in domains including massive Proper Nouns such as mathematics and biology, where it affects the accurac...

Descripción completa

Guardado en:
Detalles Bibliográficos
Autores principales: Hua Cheng, Renjie Yu, Yixin Tang, Yiquan Fang, Tao Cheng
Formato: article
Lenguaje:EN
Publicado: MDPI AG 2021
Materias:
T
Acceso en línea:https://doaj.org/article/2b40f93513304981a835803b16c74883
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
id oai:doaj.org-article:2b40f93513304981a835803b16c74883
record_format dspace
spelling oai:doaj.org-article:2b40f93513304981a835803b16c748832021-11-25T16:30:58ZText Classification Model Enhanced by Unlabeled Data for LaTeX Formula10.3390/app1122105362076-3417https://doaj.org/article/2b40f93513304981a835803b16c748832021-11-01T00:00:00Zhttps://www.mdpi.com/2076-3417/11/22/10536https://doaj.org/toc/2076-3417Generic language models pretrained on large unspecific domains are currently the foundation of NLP. Labeled data are limited in most model training due to the cost of manual annotation, especially in domains including massive Proper Nouns such as mathematics and biology, where it affects the accuracy and robustness of model prediction. However, directly applying a generic language model on a specific domain does not work well. This paper introduces a BERT-based text classification model enhanced by unlabeled data (UL-BERT) in the LaTeX formula domain. A two-stage Pretraining model based on BERT(TP-BERT) is pretrained by unlabeled data in the LaTeX formula domain. A double-prediction pseudo-labeling (DPP) method is introduced to obtain high confidence pseudo-labels for unlabeled data by self-training. Moreover, a multi-rounds teacher–student model training approach is proposed for UL-BERT model training with few labeled data and more unlabeled data with pseudo-labels. Experiments on the classification of the LaTex formula domain show that the classification accuracies have been significantly improved by UL-BERT where the F1 score has been mostly enhanced by 2.76%, and lower resources are needed in model training. It is concluded that our method may be applicable to other specific domains with enormous unlabeled data and limited labelled data.Hua ChengRenjie YuYixin TangYiquan FangTao ChengMDPI AGarticleunlabeled dataself-trainingpretrainingBERTLaTeX formulaTechnologyTEngineering (General). Civil engineering (General)TA1-2040Biology (General)QH301-705.5PhysicsQC1-999ChemistryQD1-999ENApplied Sciences, Vol 11, Iss 10536, p 10536 (2021)
institution DOAJ
collection DOAJ
language EN
topic unlabeled data
self-training
pretraining
BERT
LaTeX formula
Technology
T
Engineering (General). Civil engineering (General)
TA1-2040
Biology (General)
QH301-705.5
Physics
QC1-999
Chemistry
QD1-999
spellingShingle unlabeled data
self-training
pretraining
BERT
LaTeX formula
Technology
T
Engineering (General). Civil engineering (General)
TA1-2040
Biology (General)
QH301-705.5
Physics
QC1-999
Chemistry
QD1-999
Hua Cheng
Renjie Yu
Yixin Tang
Yiquan Fang
Tao Cheng
Text Classification Model Enhanced by Unlabeled Data for LaTeX Formula
description Generic language models pretrained on large unspecific domains are currently the foundation of NLP. Labeled data are limited in most model training due to the cost of manual annotation, especially in domains including massive Proper Nouns such as mathematics and biology, where it affects the accuracy and robustness of model prediction. However, directly applying a generic language model on a specific domain does not work well. This paper introduces a BERT-based text classification model enhanced by unlabeled data (UL-BERT) in the LaTeX formula domain. A two-stage Pretraining model based on BERT(TP-BERT) is pretrained by unlabeled data in the LaTeX formula domain. A double-prediction pseudo-labeling (DPP) method is introduced to obtain high confidence pseudo-labels for unlabeled data by self-training. Moreover, a multi-rounds teacher–student model training approach is proposed for UL-BERT model training with few labeled data and more unlabeled data with pseudo-labels. Experiments on the classification of the LaTex formula domain show that the classification accuracies have been significantly improved by UL-BERT where the F1 score has been mostly enhanced by 2.76%, and lower resources are needed in model training. It is concluded that our method may be applicable to other specific domains with enormous unlabeled data and limited labelled data.
format article
author Hua Cheng
Renjie Yu
Yixin Tang
Yiquan Fang
Tao Cheng
author_facet Hua Cheng
Renjie Yu
Yixin Tang
Yiquan Fang
Tao Cheng
author_sort Hua Cheng
title Text Classification Model Enhanced by Unlabeled Data for LaTeX Formula
title_short Text Classification Model Enhanced by Unlabeled Data for LaTeX Formula
title_full Text Classification Model Enhanced by Unlabeled Data for LaTeX Formula
title_fullStr Text Classification Model Enhanced by Unlabeled Data for LaTeX Formula
title_full_unstemmed Text Classification Model Enhanced by Unlabeled Data for LaTeX Formula
title_sort text classification model enhanced by unlabeled data for latex formula
publisher MDPI AG
publishDate 2021
url https://doaj.org/article/2b40f93513304981a835803b16c74883
work_keys_str_mv AT huacheng textclassificationmodelenhancedbyunlabeleddataforlatexformula
AT renjieyu textclassificationmodelenhancedbyunlabeleddataforlatexformula
AT yixintang textclassificationmodelenhancedbyunlabeleddataforlatexformula
AT yiquanfang textclassificationmodelenhancedbyunlabeleddataforlatexformula
AT taocheng textclassificationmodelenhancedbyunlabeleddataforlatexformula
_version_ 1718413126195478528