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 2020-12-17 14:40
Belonging Network -- Few-shot One-class Image Classification for Classes with Various Distributions --
Takumi Ohkuma, Hideki Nakayama (UT) PRMU2020-44
Abstract (in Japanese) (See Japanese page) 
(in English) Few-shot one-class image classification is the task of recognizing a particular class while rejecting test images that do not belong to the class using only a few given training images. One promising strategy for this task is to employ meta learning to transfer knowledge from existing resources. In this paper, we propose a simple meta-learning based method “Belonging Network (BeNet)”. We use and compare some basic statistics of target classes to sketch their distributions and find that simple mean and variance information of the training image set contributes to improving the performance. Despite its surprising simplicity, BeNet achieves state-of-the-art performance in our experiments.
Keyword (in Japanese) (See Japanese page) 
(in English) Few-shot learning / Image recognition / One-class classification / Meta learning / / / /  
Reference Info. IEICE Tech. Rep., vol. 120, no. 300, PRMU2020-44, pp. 36-41, Dec. 2020.
Paper # PRMU2020-44 
Date of Issue 2020-12-10 (PRMU) 
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 PRMU2020-44

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 English 
Title (in Japanese) (See Japanese page) 
Sub Title (in Japanese) (See Japanese page) 
Title (in English) Belonging Network 
Sub Title (in English) Few-shot One-class Image Classification for Classes with Various Distributions 
Keyword(1) Few-shot learning  
Keyword(2) Image recognition  
Keyword(3) One-class classification  
Keyword(4) Meta learning  
1st Author's Name Takumi Ohkuma  
1st Author's Affiliation The University of Tokyo (UT)
2nd Author's Name Hideki Nakayama  
2nd Author's Affiliation The University of Tokyo (UT)
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 ()
Date Time 2020-12-17 14:40:00 
Presentation Time 15 
Registration for PRMU 
Paper # IEICE-PRMU2020-44 
Volume (vol) IEICE-120 
Number (no) no.300 
Page pp.36-41 
#Pages IEICE-6 
Date of Issue IEICE-PRMU-2020-12-10 

[Return to Top Page]

[Return to IEICE Web Page]

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