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 2020-12-18 14:25
Zero-shot generative model considering attribute uncertainty
Yuta Sakai (Waseda Univ.), Kenta Mikawa (SIT), Masayuki Goto (Waseda Univ.) PRMU2020-59
Abstract (in Japanese) (See Japanese page) 
(in English) Classification problems in machine learning remain an important research topic. In general, classification estimates unknown labels of test data by using a training data set that consists of pairs of data features and labels representing categories. However, in general classification problems, it is not possible to estimate unkown categories that are not included in the training data set. In this research, we focus on zero-shot learning, which is a method that makes it possible to estimate categories that are not in the learning data by utilizing information (auxiliary information) that is common to the data. Among them, we pay particular attention to attribute-based zero-shot learning that utilizes attribute information as auxiliary information.
The Direct Attribute Prediction (DAP) model, which is one of the conventional methods for performing attribute-based zero-shot learning, is a discriminative model that expresses the relationship between features and categories by utilizing attributes. However, in the DAP model, estimation results other than the important attributes for category estimation may reduce the estimation accuracy of the category. Moreover, since it is a discriminative model, it may be overfitted when the number of training data is small.
Therefore, in this study, we propose an attribute-based zero-shot generation model that takes into account the uncertainty of the estimation result of the attribute information. Finally, we verify the effectiveness of the proposed method comparing the conventional method.
Keyword (in Japanese) (See Japanese page) 
(in English) Zero-shot Learning / Attribute information / generative model / classification / / / /  
Reference Info. IEICE Tech. Rep., vol. 120, no. 300, PRMU2020-59, pp. 122-127, Dec. 2020.
Paper # PRMU2020-59 
Date of Issue 2020-12-10 (PRMU) 
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 PRMU2020-59

Conference Information
Committee PRMU  
Conference Date 2020-12-17 - 2020-12-18 
Place (in Japanese) (See Japanese page) 
Place (in English) Online 
Topics (in Japanese) (See Japanese page) 
Topics (in English) Transfer learning and few shot learning 
Paper Information
Registration To PRMU 
Conference Code 2020-12-PRMU 
Language Japanese 
Title (in Japanese) (See Japanese page) 
Sub Title (in Japanese) (See Japanese page) 
Title (in English) Zero-shot generative model considering attribute uncertainty 
Sub Title (in English)  
Keyword(1) Zero-shot Learning  
Keyword(2) Attribute information  
Keyword(3) generative model  
Keyword(4) classification  
Keyword(5)  
Keyword(6)  
Keyword(7)  
Keyword(8)  
1st Author's Name Yuta Sakai  
1st Author's Affiliation Waseda University (Waseda Univ.)
2nd Author's Name Kenta Mikawa  
2nd Author's Affiliation Shonan Institute of Technology (SIT)
3rd Author's Name Masayuki Goto  
3rd Author's Affiliation Waseda University (Waseda Univ.)
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 2020-12-18 14:25:00 
Presentation Time 15 minutes 
Registration for PRMU 
Paper # PRMU2020-59 
Volume (vol) vol.120 
Number (no) no.300 
Page pp.122-127 
#Pages
Date of Issue 2020-12-10 (PRMU) 


[Return to Top Page]

[Return to IEICE Web Page]


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