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 2022-12-22 15:10
Effect of Prior Distribution when Data-Generating Process is in a Neighborhood of Singularities of Learning Machines
Nozomi Maki, Sumio Watanabe (TokyoTech) IBISML2022-46
Abstract (in Japanese) (See Japanese page) 
(in English) Learning machines which have hierarchical structure or latent variables such as deep learning or normal mixtures contains singularities in their parameter spaces, resulting that they have different learning behaviors from regular statistical models. Their generalization performance is determined by the real log canonical threshold even if the data-generating process is represented by a singularities or not. However, the generalization performance has been left unknown when a data-generating process is not equal to a singularity but in a neighborhood of a singularity. In this paper we show that the effect of the prior distribution to the generalization error when a data-generating process is not equal to a singularity but in a neighborhood of a singularity.
Keyword (in Japanese) (See Japanese page) 
(in English) Singular Learning Machines / Data-Generating Process / Prior Distribution / Generalization Error / / / /  
Reference Info. IEICE Tech. Rep., vol. 122, no. 325, IBISML2022-46, pp. 18-23, Dec. 2022.
Paper # IBISML2022-46 
Date of Issue 2022-12-15 (IBISML) 
ISSN 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 IBISML2022-46

Conference Information
Committee IBISML  
Conference Date 2022-12-22 - 2022-12-23 
Place (in Japanese) (See Japanese page) 
Place (in English) Kyoto University 
Topics (in Japanese) (See Japanese page) 
Topics (in English) Machine Learning, etc. 
Paper Information
Registration To IBISML 
Conference Code 2022-12-IBISML 
Language Japanese 
Title (in Japanese) (See Japanese page) 
Sub Title (in Japanese) (See Japanese page) 
Title (in English) Effect of Prior Distribution when Data-Generating Process is in a Neighborhood of Singularities of Learning Machines 
Sub Title (in English)  
Keyword(1) Singular Learning Machines  
Keyword(2) Data-Generating Process  
Keyword(3) Prior Distribution  
Keyword(4) Generalization Error  
Keyword(5)  
Keyword(6)  
Keyword(7)  
Keyword(8)  
1st Author's Name Nozomi Maki  
1st Author's Affiliation Tokyo Institute of Technology (TokyoTech)
2nd Author's Name Sumio Watanabe  
2nd Author's Affiliation Tokyo Institute of Technology (TokyoTech)
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 2022-12-22 15:10:00 
Presentation Time 20 minutes 
Registration for IBISML 
Paper # IBISML2022-46 
Volume (vol) vol.122 
Number (no) no.325 
Page pp.18-23 
#Pages
Date of Issue 2022-12-15 (IBISML) 


[Return to Top Page]

[Return to IEICE Web Page]


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