An application of convolutional neural networks with salient features for relation classification
Abstract Background Due to the advent of deep learning, the increasing number of studies in the biomedical domain has attracted much interest in feature extraction and classification tasks. In this research, we seek the best combination of feature set and hyperparameter setting of deep learning algo...
Main Authors: | , |
---|---|
Format: | Article |
Language: | English |
Published: |
BMC
2019-05-01
|
Series: | BMC Bioinformatics |
Subjects: | |
Online Access: | http://link.springer.com/article/10.1186/s12859-019-2808-3 |
id |
doaj-17bdf64ff7ba46ec8642b8ff6fcfd4f4 |
---|---|
record_format |
Article |
spelling |
doaj-17bdf64ff7ba46ec8642b8ff6fcfd4f42020-11-25T03:02:14ZengBMCBMC Bioinformatics1471-21052019-05-0120S1011210.1186/s12859-019-2808-3An application of convolutional neural networks with salient features for relation classificationZolzaya Dashdorj0Min Song1Department of Library and Information Science, Yonsei UniversityDepartment of Library and Information Science, Yonsei UniversityAbstract Background Due to the advent of deep learning, the increasing number of studies in the biomedical domain has attracted much interest in feature extraction and classification tasks. In this research, we seek the best combination of feature set and hyperparameter setting of deep learning algorithms for relation classification. To this end, we incorporate an entity and relation extraction tool, PKDE4J to extract biomedical features (i.e., biomedical entities, relations) for the relation classification. We compared the chosen Convolutional Neural Networks (CNN) based classification model with the most widely used learning algorithms. Results Our CNN based classification model outperforms the most widely used supervised algorithms. We achieved a significant performance on binary classification with a weighted macro-average F1-score: 94.79% using pre-extracted relevant feature combinations. For multi-class classification, the weighted macro-average F1-score is estimated around 86.95%. Conclusions Our results suggest that our proposed CNN based model using the not only single feature as the raw text of the sentences of biomedical literature, but also coupling with multiple and highlighted features extracted from the biomedical sentences could improve the classification performance significantly. We offer hyperparameter tuning and optimization approaches for our proposed model to obtain optimal hyperparameters of the models with the best performance.http://link.springer.com/article/10.1186/s12859-019-2808-3Convolutional neural networksBiomedical data analysisRelation classificationHyperparameter optimizationDeep learning |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Zolzaya Dashdorj Min Song |
spellingShingle |
Zolzaya Dashdorj Min Song An application of convolutional neural networks with salient features for relation classification BMC Bioinformatics Convolutional neural networks Biomedical data analysis Relation classification Hyperparameter optimization Deep learning |
author_facet |
Zolzaya Dashdorj Min Song |
author_sort |
Zolzaya Dashdorj |
title |
An application of convolutional neural networks with salient features for relation classification |
title_short |
An application of convolutional neural networks with salient features for relation classification |
title_full |
An application of convolutional neural networks with salient features for relation classification |
title_fullStr |
An application of convolutional neural networks with salient features for relation classification |
title_full_unstemmed |
An application of convolutional neural networks with salient features for relation classification |
title_sort |
application of convolutional neural networks with salient features for relation classification |
publisher |
BMC |
series |
BMC Bioinformatics |
issn |
1471-2105 |
publishDate |
2019-05-01 |
description |
Abstract Background Due to the advent of deep learning, the increasing number of studies in the biomedical domain has attracted much interest in feature extraction and classification tasks. In this research, we seek the best combination of feature set and hyperparameter setting of deep learning algorithms for relation classification. To this end, we incorporate an entity and relation extraction tool, PKDE4J to extract biomedical features (i.e., biomedical entities, relations) for the relation classification. We compared the chosen Convolutional Neural Networks (CNN) based classification model with the most widely used learning algorithms. Results Our CNN based classification model outperforms the most widely used supervised algorithms. We achieved a significant performance on binary classification with a weighted macro-average F1-score: 94.79% using pre-extracted relevant feature combinations. For multi-class classification, the weighted macro-average F1-score is estimated around 86.95%. Conclusions Our results suggest that our proposed CNN based model using the not only single feature as the raw text of the sentences of biomedical literature, but also coupling with multiple and highlighted features extracted from the biomedical sentences could improve the classification performance significantly. We offer hyperparameter tuning and optimization approaches for our proposed model to obtain optimal hyperparameters of the models with the best performance. |
topic |
Convolutional neural networks Biomedical data analysis Relation classification Hyperparameter optimization Deep learning |
url |
http://link.springer.com/article/10.1186/s12859-019-2808-3 |
work_keys_str_mv |
AT zolzayadashdorj anapplicationofconvolutionalneuralnetworkswithsalientfeaturesforrelationclassification AT minsong anapplicationofconvolutionalneuralnetworkswithsalientfeaturesforrelationclassification AT zolzayadashdorj applicationofconvolutionalneuralnetworkswithsalientfeaturesforrelationclassification AT minsong applicationofconvolutionalneuralnetworkswithsalientfeaturesforrelationclassification |
_version_ |
1724690682026655744 |