Home // ACHI 2021, The Fourteenth International Conference on Advances in Computer-Human Interactions // View article

Interpretation Support System for Classification Patterns Using HMM in Deep Learning with Texts

Authors:
Masayuki Ando
Yoshinobu Kawahara
Wataru Sunayama
Yuji Hatanaka

Keywords: interpretation support; deep learning, text mining; text classification; data visualization

Abstract:
This paper describes an interpretation support system for classification patterns extracted from deep learning with texts using a Hidden Markov Model (HMM), and verified its effectiveness. It is well known that classification patterns by deep learning models are often difficult to interpret the reasons derived. Therefore, an interpretation support system for deep learning classification patterns using HMMs is proposed as a tool for extracting and interpreting the learning content of deep learning. The proposed system uses an HMM to extract the contents of the learning results in deep learning with texts and provide an interface to assist in the interpretation of learned patterns. The proposed system is expected to enable system users to easily understand the complexity of deep learning, acquire new skills, and create knowledge. Verification experiments were conducted to confirm the effectiveness of the system on the basis of the learning result of deep learning classifying sentences. In the experiment, test subjects were divided into two groups. One group used the proposed system and the other used a system that displays words with high Term Frequency-Inverse Document Frequency (TFIDF) values. Both groups were instructed to provide meanings to classification patterns unusual to each output. The results show that the test subjects who used the proposed system were able to understand the meanings of the classification patterns of deep learning with texts more deeply than those who used the comparison system.

Pages: 64 to 70

Copyright: Copyright (c) IARIA, 2021

Publication date: July 18, 2021

Published in: conference

ISSN: 2308-4138

ISBN: 978-1-61208-870-9

Location: Nice, France

Dates: from July 18, 2021 to July 22, 2021