Paper Abstract and Keywords |
Presentation |
2018-06-10 10:45
Use of recurrent infomax to improve the information processing capability of input-driven recurrent neural networks Hisashi Iwade (Kyoto Univ.), Kohei Nakajima (Univ. Tokyo/JST PRESTO), Takuma Tanaka (Shiga Univ.), Toshio Aoyagi (Kyoto Univ.) NLP2018-42 CCS2018-15 |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
Reservoir computing (RC) is known to be a novel information processing framework, which exploits the inherent transient dynamics of recurrent neural networks (RNNs) as a computational resource. RC possesses high information processing capability but, in some cases, the empirical optimization using the control parameters is required. In this study, we propose recurrent infomax (RI) as a alternative method to the empirical optimization. RI is an unsupervised learning scheme that maximizes the mutual information of RNNs by adjusting the connection strengths of the network. Concequently, a delay-line structure can be observed on the network optimized by RI and the network possesses superior short-term memory, which is the ability to encode the information of the input stream in its transient dynamics. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
recurrent neural network / recurrent infomax / / / / / / |
Reference Info. |
IEICE Tech. Rep., vol. 118, no. 76, CCS2018-15, pp. 79-84, June 2018. |
Paper # |
CCS2018-15 |
Date of Issue |
2018-06-01 (NLP, CCS) |
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 |
NLP2018-42 CCS2018-15 |
Conference Information |
Committee |
NLP CCS |
Conference Date |
2018-06-08 - 2018-06-10 |
Place (in Japanese) |
(See Japanese page) |
Place (in English) |
Kyoto Terrsa |
Topics (in Japanese) |
(See Japanese page) |
Topics (in English) |
Synchronization, Networks, etc |
Paper Information |
Registration To |
CCS |
Conference Code |
2018-06-NLP-CCS |
Language |
Japanese |
Title (in Japanese) |
(See Japanese page) |
Sub Title (in Japanese) |
(See Japanese page) |
Title (in English) |
Use of recurrent infomax to improve the information processing capability of input-driven recurrent neural networks |
Sub Title (in English) |
|
Keyword(1) |
recurrent neural network |
Keyword(2) |
recurrent infomax |
Keyword(3) |
|
Keyword(4) |
|
Keyword(5) |
|
Keyword(6) |
|
Keyword(7) |
|
Keyword(8) |
|
1st Author's Name |
Hisashi Iwade |
1st Author's Affiliation |
Kyoto University (Kyoto Univ.) |
2nd Author's Name |
Kohei Nakajima |
2nd Author's Affiliation |
University of Tokyo/JST PRESTO (Univ. Tokyo/JST PRESTO) |
3rd Author's Name |
Takuma Tanaka |
3rd Author's Affiliation |
Shiga University (Shiga Univ.) |
4th Author's Name |
Toshio Aoyagi |
4th Author's Affiliation |
Kyoto University (Kyoto Univ.) |
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 |
2018-06-10 10:45:00 |
Presentation Time |
25 minutes |
Registration for |
CCS |
Paper # |
NLP2018-42, CCS2018-15 |
Volume (vol) |
vol.118 |
Number (no) |
no.75(NLP), no.76(CCS) |
Page |
pp.79-84 |
#Pages |
6 |
Date of Issue |
2018-06-01 (NLP, CCS) |
|