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 2015-06-23 14:15
Efficient sparse learning for combinatorial model by using safe screening approach
Kazuya Nakagawa, Shinya Suzumura, Masayuki Karasuyama, Ichiro Takeuchi (NIT) IBISML2015-10
Abstract (in Japanese) (See Japanese page) 
(in English) In a variety of machine learning tasks, it has been desired to incorporate high-order interaction effects of multiple covariates. However, for recent applications with a large number covariates, it is highly challenging to identify important high-order interaction features since the number of possible candidates would be extremely large.
In this paper, we propose an efficient algorithm for LASSO-based sparse learning of such high-order interaction models.
The basic strategy is to use a recently introduced safe feature screening technique by which a subset of non-active features can be identified and they can be screened-out prior to LASSO training.
However, applying safe feature screening to each of the extremely large number of high-order interaction features would be computationally infeasible.
Our key idea for solving this computational issue is to exploit the underlying tree structure among high-order interaction features.
Specifically, we introduce a set of pruning conditions of the tree such that, if one of the conditions is satis ed in a certain node, then all the high-order interaction features corresponding to its descendant nodes can be guaranteed to be non-active at the optimal solution, and they can be screened-out prior to LASSO training.
Keyword (in Japanese) (See Japanese page) 
(in English) high-order interaction model / LASSO / sparse learning / safe-screening / / / /  
Reference Info. IEICE Tech. Rep., vol. 115, no. 112, IBISML2015-10, pp. 63-68, June 2015.
Paper # IBISML2015-10 
Date of Issue 2015-06-16 (IBISML) 
ISSN Print edition: ISSN 0913-5685    Online edition: ISSN 2432-6380
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 IBISML2015-10

Conference Information
Conference Date 2015-06-23 - 2015-06-25 
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 2015-06-NC-BIO-IBISML-MPS 
Language Japanese 
Title (in Japanese) (See Japanese page) 
Sub Title (in Japanese) (See Japanese page) 
Title (in English) Efficient sparse learning for combinatorial model by using safe screening approach 
Sub Title (in English)  
Keyword(1) high-order interaction model  
Keyword(2) LASSO  
Keyword(3) sparse learning  
Keyword(4) safe-screening  
1st Author's Name Kazuya Nakagawa  
1st Author's Affiliation Nagoya Institute of Technology (NIT)
2nd Author's Name Shinya Suzumura  
2nd Author's Affiliation Nagoya Institute of Technology (NIT)
3rd Author's Name Masayuki Karasuyama  
3rd Author's Affiliation Nagoya Institute of Technology (NIT)
4th Author's Name Ichiro Takeuchi  
4th Author's Affiliation Nagoya Institute of Technology (NIT)
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 2015-06-23 14:15:00 
Presentation Time 25 minutes 
Registration for IBISML 
Paper # IBISML2015-10 
Volume (vol) vol.115 
Number (no) no.112 
Page pp.63-68 
Date of Issue 2015-06-16 (IBISML) 

[Return to Top Page]

[Return to IEICE Web Page]

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