IEICE Technical Committee Submission System
Conference Paper's Information
Online Proceedings
[Sign in]
Tech. Rep. Archives
 Go Top Page Go Previous   [Japanese] / [English] 

Paper Abstract and Keywords
Presentation 2019-02-28 13:55
Model Compression for ECG Signals Outlier Detection Hardware trained by Sparse Robust Deep Autoencoder
Naoto Soga, Shimpei Sato, Hiroki Nakahara (Titech) VLD2018-114 HWS2018-77
Abstract (in Japanese) (See Japanese page) 
(in English) In recent years, portable electrocardiographs and wearable devices have begun to spread so that electrocar- diogram (ECG) signals can be recorded in everyday life. Many types of research have been conducted to use machine learning techniques, including deep learning techniques, to analyze ECG data. However, deep learning models often have too many parameters to implement on mobile hardware. In this research, we propose a method to implement an ECG outlier detector using an autoencoder, which is based on a neural network, in a small built-in device. As a learning method, Robust Deep Autoencoder, one of the unsupervised learning method, was used. A sparseness technique was applied to the autoencoder, and the number of parameters was reduced. Also, Weight Sharing was applied to the obtained weight parameters. With Weight Sharing, the capacity occupied by the weight parameters was reduced by 75% compared with the case where only the sparseness technique was applied. We implemented the obtained Autoencoder on an FPGA. The speed is 20.2 times faster, and 182 times more power efficient than CPUs.
Keyword (in Japanese) (See Japanese page) 
(in English) outlier detection / autoencoder / sparse network / K-Means / unsupervised learning / FPGA / /  
Reference Info. IEICE Tech. Rep., vol. 118, no. 457, VLD2018-114, pp. 127-132, Feb. 2019.
Paper # VLD2018-114 
Date of Issue 2019-02-20 (VLD, HWS) 
ISSN Print edition: ISSN 0913-5685    Online edition: ISSN 2432-6380
Copyright
and
reproduction
All rights are reserved and no part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Notwithstanding, instructors are permitted to photocopy isolated articles for noncommercial classroom use without fee. (License No.: 10GA0019/12GB0052/13GB0056/17GB0034/18GB0034)
Download PDF VLD2018-114 HWS2018-77

Conference Information
Committee HWS VLD  
Conference Date 2019-02-27 - 2019-03-02 
Place (in Japanese) (See Japanese page) 
Place (in English) Okinawa Ken Seinen Kaikan 
Topics (in Japanese) (See Japanese page) 
Topics (in English) Design Technology for System-on-Silicon, Hardware Security, etc. 
Paper Information
Registration To VLD 
Conference Code 2019-02-HWS-VLD 
Language Japanese 
Title (in Japanese) (See Japanese page) 
Sub Title (in Japanese) (See Japanese page) 
Title (in English) Model Compression for ECG Signals Outlier Detection Hardware trained by Sparse Robust Deep Autoencoder 
Sub Title (in English)  
Keyword(1) outlier detection  
Keyword(2) autoencoder  
Keyword(3) sparse network  
Keyword(4) K-Means  
Keyword(5) unsupervised learning  
Keyword(6) FPGA  
Keyword(7)  
Keyword(8)  
1st Author's Name Naoto Soga  
1st Author's Affiliation Tokyo Institute of Technology (Titech)
2nd Author's Name Shimpei Sato  
2nd Author's Affiliation Tokyo Institute of Technology (Titech)
3rd Author's Name Hiroki Nakahara  
3rd Author's Affiliation Tokyo Institute of Technology (Titech)
4th Author's Name  
4th Author's Affiliation ()
5th Author's Name  
5th Author's Affiliation ()
6th Author's Name  
6th Author's Affiliation ()
7th Author's Name  
7th Author's Affiliation ()
8th Author's Name  
8th Author's Affiliation ()
9th Author's Name  
9th Author's Affiliation ()
10th Author's Name  
10th Author's Affiliation ()
11th Author's Name  
11th Author's Affiliation ()
12th Author's Name  
12th Author's Affiliation ()
13th Author's Name  
13th Author's Affiliation ()
14th Author's Name  
14th Author's Affiliation ()
15th Author's Name  
15th Author's Affiliation ()
16th Author's Name  
16th Author's Affiliation ()
17th Author's Name  
17th Author's Affiliation ()
18th Author's Name  
18th Author's Affiliation ()
19th Author's Name  
19th Author's Affiliation ()
20th Author's Name  
20th Author's Affiliation ()
Speaker Author-1 
Date Time 2019-02-28 13:55:00 
Presentation Time 25 minutes 
Registration for VLD 
Paper # VLD2018-114, HWS2018-77 
Volume (vol) vol.118 
Number (no) no.457(VLD), no.458(HWS) 
Page pp.127-132 
#Pages
Date of Issue 2019-02-20 (VLD, HWS) 


[Return to Top Page]

[Return to IEICE Web Page]


The Institute of Electronics, Information and Communication Engineers (IEICE), Japan