ETRI-Knowledge Sharing Plaform

KOREAN
논문 검색
Type SCI
Year ~ Keyword

Detail

Conference Paper Arrhythmia Detection Using Convolutional Neural Networks with Temporal Attention Mechanism
Cited 1 time in scopus Share share facebook twitter linkedin kakaostory
Authors
Muhammad Zubair, Sungpil Woo, Sunhwan Lim, Chan-Won Park
Issue Date
2021-10
Citation
International Conference on Information and Communication Technology Convergence (ICTC) 2021, pp.1101-1103
Publisher
IEEE
Language
English
Type
Conference Paper
DOI
https://dx.doi.org/10.1109/ICTC52510.2021.9620980
Abstract
This paper presents a deep learning approach for arrhythmia detection using convolutional neural networks with attention mechanism. In order to learn an efficient deep feature representation from raw ECG data, we proposed a temporal attention module that extract target-specific information. The proposed two-branch attention module aims to extract short-term and long-term morphological patterns of the beats with different convolutional kernels. We also followed the recommendations of the Association for the Advancement of Medical Instrumentation (AAMI) for beat labeling. The experimental results reveal that the proposed approach significantly discriminate between different beats types without using hand crafted features and achieve an improved classification performance as compared to other methods.
KSP Keywords
Attention mechanism, Classification Performance, Convolution neural network(CNN), Learning approach, Medical instrumentation, Morphological patterns, arrhythmia detection, deep feature representation, deep learning(DL), neural network(NN), short-term