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 2018-01-23 11:15
Retraining anomaly detection model using Autoencoder
Yasuhiro Ikeda, Keisuke Ishibashi, Yusuke Nakano, Keishiro Watanabe, Ryoichi Kawahara (NTT) IN2017-84
Abstract (in Japanese) (See Japanese page) 
(in English) An autoencoder has been attracting much attention as an anomaly detection algorithm.
The autoencoder enables unsupervised learning by using input data as output labels,
and therefore by training the autoencoder with data in normal time,
it is trained to output abnormality of test data
according to how far they are different from the training data.
The autoencoder therefore seems to be desireble as an anomaly detection algorithm
under the situation that abnormal data cannot be obtained sufficiently.
However, since the ``normal state'' of systems will not be static
and false positives due to insufficient training may be unavoidable,
retraining the model according to the data trend and the false-positive detection
is required for continually using the autoencoder for anomaly detection.
In this paper, we propose a retraining algorithm of the autoencoder
including the retraining of ``normal outlier'' which is often a problem in system surveillance
due to temporal high load, for example,
and also evaluate the algorithm through network benchmark data.
Keyword (in Japanese) (See Japanese page) 
(in English) Deep Learning / Autoencoder / Retraining / / / / /  
Reference Info. IEICE Tech. Rep., vol. 117, no. 397, IN2017-84, pp. 77-82, Jan. 2018.
Paper # IN2017-84 
Date of Issue 2018-01-15 (IN) 
ISSN Print edition: ISSN 0913-5685  Online edition: ISSN 2432-6380
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 IN2017-84

Conference Information
Committee IN  
Conference Date 2018-01-22 - 2018-01-23 
Place (in Japanese) (See Japanese page) 
Place (in English) WINC AICHI 
Topics (in Japanese) (See Japanese page) 
Topics (in English) Contents Distribution, Social Networking Services, Data Analytics and Processing Platform, Big data, etc. 
Paper Information
Registration To IN 
Conference Code 2018-01-IN 
Language Japanese 
Title (in Japanese) (See Japanese page) 
Sub Title (in Japanese) (See Japanese page) 
Title (in English) Retraining anomaly detection model using Autoencoder 
Sub Title (in English)  
Keyword(1) Deep Learning  
Keyword(2) Autoencoder  
Keyword(3) Retraining  
1st Author's Name Yasuhiro Ikeda  
1st Author's Affiliation NTT (NTT)
2nd Author's Name Keisuke Ishibashi  
2nd Author's Affiliation NTT (NTT)
3rd Author's Name Yusuke Nakano  
3rd Author's Affiliation NTT (NTT)
4th Author's Name Keishiro Watanabe  
4th Author's Affiliation NTT (NTT)
5th Author's Name Ryoichi Kawahara  
5th Author's Affiliation NTT (NTT)
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 ()
Date Time 2018-01-23 11:15:00 
Presentation Time 25 
Registration for IN 
Paper # IEICE-IN2017-84 
Volume (vol) IEICE-117 
Number (no) no.397 
Page pp.77-82 
#Pages IEICE-6 
Date of Issue IEICE-IN-2018-01-15 

[Return to Top Page]

[Return to IEICE Web Page]

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