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 2017-03-07 14:00
CTC network with explicit representation vector of Markov property
Yuta Kawachi, Taichi Asami, Yoshikazu Yamaguchi, Yushi Aono (NTT) IBISML2016-111
Abstract (in Japanese) (See Japanese page) 
(in English) Current neural acoustic models are incapable of utilizing language resources except speech transcriptions. So toward the realization of true neural end-to-end speech recognition, this property is problematic. In this paper, we present a method for general sequence classification task based on connectionist temporal classification with a structure keeping the previously output symbols and a matrix explicitly represents bigram properties. In our experiment, we defined a problem easily solved by using Markov properties of output sequences. Finally we could obtain interpretable neural network parameters and accuracy improvements from the baseline LSTM-CTC model.
Keyword (in Japanese) (See Japanese page) 
(in English) temporal classification / deep learning / CTC / end-to-end / / / /  
Reference Info. IEICE Tech. Rep., vol. 116, no. 500, IBISML2016-111, pp. 83-88, March 2017.
Paper # IBISML2016-111 
Date of Issue 2017-02-27 (IBISML) 
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 IBISML2016-111

Conference Information
Committee IBISML  
Conference Date 2017-03-06 - 2017-03-07 
Place (in Japanese) (See Japanese page) 
Place (in English) Tokyo Institute of Technology 
Topics (in Japanese) (See Japanese page) 
Topics (in English) Statistical Mathematics, Machine Learning, Data Mining, etc. 
Paper Information
Registration To IBISML 
Conference Code 2017-03-IBISML 
Language Japanese 
Title (in Japanese) (See Japanese page) 
Sub Title (in Japanese) (See Japanese page) 
Title (in English) CTC network with explicit representation vector of Markov property 
Sub Title (in English)  
Keyword(1) temporal classification  
Keyword(2) deep learning  
Keyword(3) CTC  
Keyword(4) end-to-end  
Keyword(5)  
Keyword(6)  
Keyword(7)  
Keyword(8)  
1st Author's Name Yuta Kawachi  
1st Author's Affiliation NTT MedNTT Corporationia Intelligence Laboratories (NTT)
2nd Author's Name Taichi Asami  
2nd Author's Affiliation NTT MedNTT Corporationia Intelligence Laboratories (NTT)
3rd Author's Name Yoshikazu Yamaguchi  
3rd Author's Affiliation NTT MediaNTT Corporation Intelligence Laboratories (NTT)
4th Author's Name Yushi Aono  
4th Author's Affiliation NTT Corporation (NTT)
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 2017-03-07 14:00:00 
Presentation Time 30 minutes 
Registration for IBISML 
Paper # IBISML2016-111 
Volume (vol) vol.116 
Number (no) no.500 
Page pp.83-88 
#Pages
Date of Issue 2017-02-27 (IBISML) 


[Return to Top Page]

[Return to IEICE Web Page]


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