KSC 2020
Current Result Document : 2 / 2
ÇѱÛÁ¦¸ñ(Korean Title) |
µ¥ÀÌÅÍ Áõ° ±â¼úÀ» È°¿ëÇÑ ECG ºÐ·ù±â ¼º´É °³¼± |
¿µ¹®Á¦¸ñ(English Title) |
Improving the performance of an ECG classifier based on data augmentation technology |
ÀúÀÚ(Author) |
ÁÖ¿ì
¹ÚÁÖ¿µ
¾ÈÁø¼º
°°æÅÂ
Yu Zhou
Juyoung Park
Jisung An
Kyungtae Kang
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 47 NO. 02 PP. 0690 ~ 0692 (2020. 12) |
Çѱ۳»¿ë (Korean Abstract) |
|
¿µ¹®³»¿ë (English Abstract) |
Arrhythmia is most commonly diagnosed using an electrocardiogram (ECG) with classification techniques based on various heart arrhythmia characteristics, The most commonly used heartbeat data set is the MIT-BIH arrhythmia database. A heartbeat is classified into normal or abnormal types, of which abnormal types can be classified into 15 types. However, despite many studies on prior classification techniques, these 15 types of data are still not uniformly classified and distributed. In this paper, we propose the application of a generative adversarial network, a data augmentation technique, in the arrhythmia diagnosis process and enhance the classification of the 15 abnormal heartbeat types. We expect to achieve accurate classification of normal and abnormal types of heartbeats through our study and the further classification of abnormal heartbeats into specific types.
|
Å°¿öµå(Keyword) |
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|