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 2017-11-10 13:00
Safe Screening for Large Margin Metric Learning
Tomoki Yoshida (NITech), Ichiro Takeuchi (NITech/NIMS/RIKEN), Masayuki Karasuyama (NITech/NIMS/JST) IBISML2017-64
Abstract (in Japanese) (See Japanese page) 
(in English) Large margin metric learning learns the optimal Mahalanobis distance for classification problem based on the margin maximization principle. The loss function in this method is defined for triples of instances, called triplet. However, the number of triplets can be huge even for small datasets and it causes large computational cost for the optimization. In this paper, we propose safe triplet screening by which unnecessary triplets can be safely eliminated before or during learning without loosing the optimality. We analyze several types of screening rules derived by considering a region (bound) containing the optimal solution, and evaluate effectiveness of the proposed method by experiments on several benchmark datasets.
Keyword (in Japanese) (See Japanese page) 
(in English) Mahalanobis distance / Safe Screening / Metric Learning / / / / /  
Reference Info. IEICE Tech. Rep., vol. 117, no. 293, IBISML2017-64, pp. 219-226, Nov. 2017.
Paper # IBISML2017-64 
Date of Issue 2017-11-02 (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 IBISML2017-64

Conference Information
Committee IBISML  
Conference Date 2017-11-08 - 2017-11-10 
Place (in Japanese) (See Japanese page) 
Place (in English) Univ. of Tokyo 
Topics (in Japanese) (See Japanese page) 
Topics (in English) Information-Based Induction Science Workshop (IBIS2017) 
Paper Information
Registration To IBISML 
Conference Code 2017-11-IBISML 
Language Japanese 
Title (in Japanese) (See Japanese page) 
Sub Title (in Japanese) (See Japanese page) 
Title (in English) Safe Screening for Large Margin Metric Learning 
Sub Title (in English)  
Keyword(1) Mahalanobis distance  
Keyword(2) Safe Screening  
Keyword(3) Metric Learning  
Keyword(4)  
Keyword(5)  
Keyword(6)  
Keyword(7)  
Keyword(8)  
1st Author's Name Tomoki Yoshida  
1st Author's Affiliation Nagoya Institute of Technology (NITech)
2nd Author's Name Ichiro Takeuchi  
2nd Author's Affiliation Nagoya Institute of Technology/National Institute for Material Science/Institute of Physical and Chemical Research (NITech/NIMS/RIKEN)
3rd Author's Name Masayuki Karasuyama  
3rd Author's Affiliation Nagoya Institute of Technology/National Institute for Material Science/Japan Science and Technology Agency (NITech/NIMS/JST)
4th Author's Name  
4th Author's Affiliation ()
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 2017-11-10 13:00:00 
Presentation Time 150 minutes 
Registration for IBISML 
Paper # IBISML2017-64 
Volume (vol) vol.117 
Number (no) no.293 
Page pp.219-226 
#Pages
Date of Issue 2017-11-02 (IBISML) 


[Return to Top Page]

[Return to IEICE Web Page]


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