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 2022-06-29 13:55
Optimization of recurrent neural network structure by controlling symmetry of weight matrix
Arisa Fujimoto, Hideaki Yamamoto, Satoshi Moriya (Tohoku Univ.), Keita Tokuda (Tsukuba Univ.), Yuichi Katori (Future Univ. Hakodate), Shigeo Sato (Tohoku Univ.) NC2022-26 IBISML2022-26
Abstract (in Japanese) (See Japanese page) 
(in English) In this study, we investigated the relationship between the strength of symmetry in a Gaussian random matrix of a recurrent neural network (RNN) and its reservoir computing performance for optimizing network topology of RNNs in reservoir computing models. The performance of the random network with various symmetry was evaluated based on a task that outputs time-series signals for writing handwritten digits. Matrix symmetry influenced the speed of the dynamics in the reservoir layer, and the highest performance was achieved when the speed of the target output was comparable to that of the network dynamics. This result suggests that, in addition to the conventional adjustment of the spectral radius, optimization of the matrix symmetry in the weight matrix could be used to efficiently improve the performance of reservoir networks.
Keyword (in Japanese) (See Japanese page) 
(in English) Recurrent neural network / Reservoir computing / Weight matrix / Symmetry / Eigenvalue / / /  
Reference Info. IEICE Tech. Rep., vol. 122, no. 89, NC2022-26, pp. 184-188, June 2022.
Paper # NC2022-26 
Date of Issue 2022-06-20 (NC, IBISML) 
ISSN 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 NC2022-26 IBISML2022-26

Conference Information
Committee NC IBISML IPSJ-BIO IPSJ-MPS  
Conference Date 2022-06-27 - 2022-06-29 
Place (in Japanese) (See Japanese page) 
Place (in English)  
Topics (in Japanese) (See Japanese page) 
Topics (in English)  
Paper Information
Registration To NC 
Conference Code 2022-06-NC-IBISML-BIO-MPS 
Language Japanese 
Title (in Japanese) (See Japanese page) 
Sub Title (in Japanese) (See Japanese page) 
Title (in English) Optimization of recurrent neural network structure by controlling symmetry of weight matrix 
Sub Title (in English)  
Keyword(1) Recurrent neural network  
Keyword(2) Reservoir computing  
Keyword(3) Weight matrix  
Keyword(4) Symmetry  
Keyword(5) Eigenvalue  
Keyword(6)  
Keyword(7)  
Keyword(8)  
1st Author's Name Arisa Fujimoto  
1st Author's Affiliation Tohoku University (Tohoku Univ.)
2nd Author's Name Hideaki Yamamoto  
2nd Author's Affiliation Tohoku University (Tohoku Univ.)
3rd Author's Name Satoshi Moriya  
3rd Author's Affiliation Tohoku University (Tohoku Univ.)
4th Author's Name Keita Tokuda  
4th Author's Affiliation Tsukuba University (Tsukuba Univ.)
5th Author's Name Yuichi Katori  
5th Author's Affiliation Future Univeisity Hakodate (Future Univ. Hakodate)
6th Author's Name Shigeo Sato  
6th Author's Affiliation Tohoku University (Tohoku Univ.)
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 2022-06-29 13:55:00 
Presentation Time 25 minutes 
Registration for NC 
Paper # NC2022-26, IBISML2022-26 
Volume (vol) vol.122 
Number (no) no.89(NC), no.90(IBISML) 
Page pp.184-188 
#Pages 5 
Date of Issue 2022-06-20 (NC, IBISML) 

[Return to Top Page]

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