Information: Join today and make your research activities more affordable! Technical workshop participation fees and annual registration fees are available at member rates.
Notice: [Important] Announcement of Changes to Registration Fee Payment and Manuscript Upload Procedures for IEICE Technical Meetings
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 2014-03-14 15:30
Experimental study on effect of pre-training in deep learning through visualization of unit outputs
Tsubasa Ochiai (Doshisha Univ./NICT), Hideyuki Watanabe (NICT), Shigeru Katagiri, Miho Ohsaki (Doshisha Univ.), Shigeki Matsuda, Chiori Hori (NICT) PRMU2013-210
Abstract (in Japanese) (See Japanese page) 
(in English) To clarify the capability of recent powerful classifier concept, Deep Neural Networks (DNN), we experimentally
investigate effects of the pre-training used to initialize DNN. A deep neural network is first pre-trained using
Restricted Boltzmann Machine (RBM), then it is run as an embodiment of Deep Belief Networks, which basically
possess associative memory function, and a Deep Autoencoder, which is expected to realize feature representation
for an input pattern over the inner layers of network. Analyses are conducted through the visualization of network
unit outputs. Based on the experiments, we reveal that the RBM-based pre-training successfully makes networks
memorize some information of training patterns and also represent pattern features inside the networks.
Keyword (in Japanese) (See Japanese page) 
(in English) Deep Learning / Pre-training / Visualization of unit outputs / Deep Neural Networks / / / /  
Reference Info. IEICE Tech. Rep., vol. 113, no. 493, PRMU2013-210, pp. 253-258, March 2014.
Paper # PRMU2013-210 
Date of Issue 2014-03-06 (PRMU) 
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 PRMU2013-210

Conference Information
Committee PRMU  
Conference Date 2014-03-13 - 2014-03-14 
Place (in Japanese) (See Japanese page) 
Place (in English)  
Topics (in Japanese) (See Japanese page) 
Topics (in English)  
Paper Information
Registration To PRMU 
Conference Code 2014-03-PRMU 
Language Japanese 
Title (in Japanese) (See Japanese page) 
Sub Title (in Japanese) (See Japanese page) 
Title (in English) Experimental study on effect of pre-training in deep learning through visualization of unit outputs 
Sub Title (in English)  
Keyword(1) Deep Learning  
Keyword(2) Pre-training  
Keyword(3) Visualization of unit outputs  
Keyword(4) Deep Neural Networks  
Keyword(5)  
Keyword(6)  
Keyword(7)  
Keyword(8)  
1st Author's Name Tsubasa Ochiai  
1st Author's Affiliation Doshisha University/National Institute of Information and Communications Technology (Doshisha Univ./NICT)
2nd Author's Name Hideyuki Watanabe  
2nd Author's Affiliation National Institute of Information and Communications Technology (NICT)
3rd Author's Name Shigeru Katagiri  
3rd Author's Affiliation Doshisha University (Doshisha Univ.)
4th Author's Name Miho Ohsaki  
4th Author's Affiliation Doshisha University (Doshisha Univ.)
5th Author's Name Shigeki Matsuda  
5th Author's Affiliation National Institute of Information and Communications Technology (NICT)
6th Author's Name Chiori Hori  
6th Author's Affiliation National Institute of Information and Communications Technology (NICT)
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 ()
21st Author's Name  
21st Author's Affiliation ()
22nd Author's Name  
22nd Author's Affiliation ()
23rd Author's Name  
23rd Author's Affiliation ()
24th Author's Name  
24th Author's Affiliation ()
25th Author's Name  
25th Author's Affiliation ()
26th Author's Name / /
26th Author's Affiliation ()
()
27th Author's Name / /
27th Author's Affiliation ()
()
28th Author's Name / /
28th Author's Affiliation ()
()
29th Author's Name / /
29th Author's Affiliation ()
()
30th Author's Name / /
30th Author's Affiliation ()
()
31st Author's Name / /
31st Author's Affiliation ()
()
32nd Author's Name / /
32nd Author's Affiliation ()
()
33rd Author's Name / /
33rd Author's Affiliation ()
()
34th Author's Name / /
34th Author's Affiliation ()
()
35th Author's Name / /
35th Author's Affiliation ()
()
36th Author's Name / /
36th Author's Affiliation ()
()
Speaker Author-1 
Date Time 2014-03-14 15:30:00 
Presentation Time 30 minutes 
Registration for PRMU 
Paper # PRMU2013-210 
Volume (vol) vol.113 
Number (no) no.493 
Page pp.253-258 
#Pages 6 
Date of Issue 2014-03-06 (PRMU) 

[Return to Top Page]

[Return to IEICE Web Page]
The Institute of Electronics, Information and Communication Engineers (IEICE), Japan