Performance evaluation of deep learning models

Myeonghwi Kim, Beakcheol Jang

Research output: Contribution to journalArticlepeer-review

Abstract

Text classification involves the process of insights mining from unstructured, textual data and natural language documents. Text based sentiment classification has recently gained a huge research audience, mainly due to the advances in natural language processing. However, few studies have tried to compare the performance of various learning methods involved when facing a similar computing task. In this paper, we experimentally compare the mostly used models in text analytics which are multilayer perceptron (MLP), convolutional neural networks (CNN s), long short term memory (LSTM) and Bi-LSTM models. We perform binary sentiment classification using collected news data, and the models were evaluated in terms of the accuracy. As a result, Bi-LSTM showed the best performance, MLP showed constant performance regardless of data set, and CNN and LSTM showed good performance as data size increased. LSTM provides better performance on more complex data.

Original languageEnglish
Pages (from-to)83-90
Number of pages8
JournalJP Journal of Heat and Mass Transfer
Volume2020
Issue numberSpecial Issue 1
DOIs
Publication statusPublished - 2020

Bibliographical note

Publisher Copyright:
© 2020 Pushpa Publishing House, Prayagraj, India.

All Science Journal Classification (ASJC) codes

  • Atomic and Molecular Physics, and Optics

Fingerprint Dive into the research topics of 'Performance evaluation of deep learning models'. Together they form a unique fingerprint.

Cite this