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 2017-06-25 10:20
Learning with linearly transformed l0 sparsity
Naoki Marumo, Tomoharu Iwata (NTT) IBISML2017-8
Abstract (in Japanese) (See Japanese page) 
(in English) We consider a class of non-convex optimization problems with linearly transformed
sparsity constraints, which includes a wide range of problems in machine learning.
Since this problem is difficult to solve directly, $ell_1$ relaxation methods
have been extensively used.
However, $ell_1$ methods often suffer from large estimation bias.
To avoid this problem, we propose a provable algorithm for solving the $ell_0$ problem without relaxation.
Our analysis shows that the proposed method converges linearly to an approximately global optimum under mild assumptions.
With our experiments, we confirm that the proposed method is faster and more accurate than existing $ell_1$ and $ell_0$ methods and that the proposed method can capture the true sparsity level more correctly.
Furthermore, the results show that the proposed method works well under severe conditions such as small sample size and high dimension.
Keyword (in Japanese) (See Japanese page) 
(in English) nonconvex optimization / sparse optimization / machine learning / / / / /  
Reference Info. IEICE Tech. Rep., vol. 117, no. 110, IBISML2017-8, pp. 193-199, June 2017.
Paper # IBISML2017-8 
Date of Issue 2017-06-17 (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 IBISML2017-8

Conference Information
Committee NC IPSJ-BIO IBISML IPSJ-MPS  
Conference Date 2017-06-23 - 2017-06-25 
Place (in Japanese) (See Japanese page) 
Place (in English) Okinawa Institute of Science and Technology 
Topics (in Japanese) (See Japanese page) 
Topics (in English) Machine Learning Approach to Biodata Mining, and General 
Paper Information
Registration To IBISML 
Conference Code 2017-06-NC-BIO-IBISML-MPS 
Language Japanese 
Title (in Japanese) (See Japanese page) 
Sub Title (in Japanese) (See Japanese page) 
Title (in English) Learning with linearly transformed l0 sparsity 
Sub Title (in English)  
Keyword(1) nonconvex optimization  
Keyword(2) sparse optimization  
Keyword(3) machine learning  
Keyword(4)  
Keyword(5)  
Keyword(6)  
Keyword(7)  
Keyword(8)  
1st Author's Name Naoki Marumo  
1st Author's Affiliation Nippon Telegraph and Telephone (NTT)
2nd Author's Name Tomoharu Iwata  
2nd Author's Affiliation Nippon Telegraph and Telephone (NTT)
3rd Author's Name  
3rd Author's Affiliation ()
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 2017-06-25 10:20:00 
Presentation Time 25 minutes 
Registration for IBISML 
Paper # IBISML2017-8 
Volume (vol) vol.117 
Number (no) no.110 
Page pp.193-199 
#Pages
Date of Issue 2017-06-17 (IBISML) 


[Return to Top Page]

[Return to IEICE Web Page]


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