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 2014-11-18 15:00
[Poster Presentation] Optimization Method of LASSO Hyperparameter using WAIC
Dai Miyazaki, Sumio Watanabe (Tokyo Tech) IBISML2014-63
Abstract (in Japanese) (See Japanese page) 
(in English) LASSO (Least Absolute Shrinkage and Selection Operator) was proposed as a regression method using a penalty term made of the sum of absolute values of parameters, which is expected to be effective when the true parameter is sparse. However, it is not yet established how to optimize its hyperparameter. In this paper, we study LASSO from the Bayesian point of view, and propose an optimization method of the hyperparameter using the information criterion WAIC which is an asymptotic unbiased estimator of the generalization error. Experimentally, we show four points. (1) By using WAIC, the average generalization error of LASSO can be estimated. (2) If the true parameter is sparse, then LASSO is effective and the optimal hyperparameter can be found by WAIC. (3) If the true parameter is not sparse, then LASSO is not effective which can be observed by using WAIC. (4) In, LASSO, the hyperparameter which minimizes the marginal likelihood is different from the hyperparameter which minimizes WAIC and the average generalization error.
Keyword (in Japanese) (See Japanese page) 
(in English) LASSO regression / Hyperparameter / WAIC / / / / /  
Reference Info. IEICE Tech. Rep., vol. 114, no. 306, IBISML2014-63, pp. 213-218, Nov. 2014.
Paper # IBISML2014-63 
Date of Issue 2014-11-10 (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 IBISML2014-63

Conference Information
Committee IBISML  
Conference Date 2014-11-17 - 2014-11-19 
Place (in Japanese) (See Japanese page) 
Place (in English) Nagoya Univ. 
Topics (in Japanese) (See Japanese page) 
Topics (in English)  
Paper Information
Registration To IBISML 
Conference Code 2014-11-IBISML 
Language Japanese 
Title (in Japanese) (See Japanese page) 
Sub Title (in Japanese) (See Japanese page) 
Title (in English) Optimization Method of LASSO Hyperparameter using WAIC 
Sub Title (in English)  
Keyword(1) LASSO regression  
Keyword(2) Hyperparameter  
Keyword(3) WAIC  
Keyword(4)  
Keyword(5)  
Keyword(6)  
Keyword(7)  
Keyword(8)  
1st Author's Name Dai Miyazaki  
1st Author's Affiliation Tokyo Institute of Technology (Tokyo Tech)
2nd Author's Name Sumio Watanabe  
2nd Author's Affiliation Tokyo Institute of Technology (Tokyo Tech)
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 2014-11-18 15:00:00 
Presentation Time 180 minutes 
Registration for IBISML 
Paper # IBISML2014-63 
Volume (vol) vol.114 
Number (no) no.306 
Page pp.213-218 
#Pages
Date of Issue 2014-11-10 (IBISML) 


[Return to Top Page]

[Return to IEICE Web Page]


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