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 2015-12-02 13:55
Automation of high performance system building for large vocabulary speech recognition using evolution strategy with pareto optimality
Takafumi Moriya, Tomohiro Tanaka, Takahiro Shinozaki (Tokyo Tech), Shinji Watanabe (MERL), Kevin Duh (NAIST) SP2015-75
Abstract (in Japanese) (See Japanese page) 
(in English) The performance of speech recognition tasks can be significantly improved by the use of deep neural networks (DNN). However, when building a high performance speech recognition system, the laborious effort required by human experts in tuning numerous parameters remains a prominent obstacle. In addition, computation time can be prohibitive when training large DNN models. The goal of this paper is to automate the process. We propose to tune DNN-HMM based large vocabulary speech recognition systems using the covariance matrix adaptation evolution strategy (CMA-ES) with a multi-objective Pareto optimization. This optimizes systems to achieve both high-accuracy and compact model size. Compared to a strong manually-tuned configuration borrowed from a similar system, our approach automatically discovered systems with lower WER by 0.48%, and systems with 59% smaller model size while keeping WER constant. The optimized training script is released in the Kaldi speech recognition toolkit as the first publicly available recipe for Japanese large vocabulary speech recognition.
Keyword (in Japanese) (See Japanese page) 
(in English) large vocabulary speech recognition / evolution strategy / deep neural network / multi-objective optimization / / / /  
Reference Info. IEICE Tech. Rep., vol. 115, no. 346, SP2015-75, pp. 31-36, Dec. 2015.
Paper # SP2015-75 
Date of Issue 2015-11-25 (SP) 
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)
Notes on Review This article is a technical report without peer review, and its polished version will be published elsewhere.
Download PDF SP2015-75

Conference Information
Committee NLC IPSJ-NL SP IPSJ-SLP  
Conference Date 2015-12-02 - 2015-12-04 
Place (in Japanese) (See Japanese page) 
Place (in English) Nagoya Inst of Tech. 
Topics (in Japanese) (See Japanese page) 
Topics (in English) The Second Natural Language Processing Symposium & The 17th Spoken Language Symposium 
Paper Information
Registration To SP 
Conference Code 2015-12-NLC-NL-SP-SLP 
Language Japanese 
Title (in Japanese) (See Japanese page) 
Sub Title (in Japanese) (See Japanese page) 
Title (in English) Automation of high performance system building for large vocabulary speech recognition using evolution strategy with pareto optimality 
Sub Title (in English)  
Keyword(1) large vocabulary speech recognition  
Keyword(2) evolution strategy  
Keyword(3) deep neural network  
Keyword(4) multi-objective optimization  
Keyword(5)  
Keyword(6)  
Keyword(7)  
Keyword(8)  
1st Author's Name Takafumi Moriya  
1st Author's Affiliation Tokyo Institute of Technology (Tokyo Tech)
2nd Author's Name Tomohiro Tanaka  
2nd Author's Affiliation Tokyo Institute of Technology (Tokyo Tech)
3rd Author's Name Takahiro Shinozaki  
3rd Author's Affiliation Tokyo Institute of Technology (Tokyo Tech)
4th Author's Name Shinji Watanabe  
4th Author's Affiliation Mitsubishi Electric Research Laboratories (MERL)
5th Author's Name Kevin Duh  
5th Author's Affiliation Nara Institute of Science and Technology (NAIST)
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 ()
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 2015-12-02 13:55:00 
Presentation Time 25 minutes 
Registration for SP 
Paper # SP2015-75 
Volume (vol) vol.115 
Number (no) no.346 
Page pp.31-36 
#Pages 6 
Date of Issue 2015-11-25 (SP) 

[Return to Top Page]

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