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 2020-09-02 16:30
Representation Learning using Video Frame Prediction and Contrastive Learning
Hidemoto Nakada, Hideki Asoh (AIST) PRMU2020-17
Abstract (in Japanese) (See Japanese page) 
(in English) The recent development in the unsupervised learning area enabled accuracy in the downstream tasks that equal the one with supervised pretraining using the labeled data, with transfer learning using a small amount of labeled data.
Most of these unsupervised methods rely on the data augmentation techniques, such as random cropping, known to be empirically effective in giving the models' induction bias.
We propose to use the inherent continuity of video frames instead of empirical data augmentation techniques.
Namely, we propose two methods, 1) use another frame in the same video clip as a positive sample, 2) predict a frame using other frames in the same clip and use the predicted frame as a positive sample.
We evaluated these methods with image classification as the downstream task. The result showed that while the former method could not achieve good accuracy, the second method equals
the traditional data augmentation techniques.
Keyword (in Japanese) (See Japanese page) 
(in English) Representation Learning / Contrastive Learning / Unsupervised Learning / / / / /  
Reference Info. IEICE Tech. Rep., vol. 120, no. 154, PRMU2020-17, pp. 59-64, Sept. 2020.
Paper # PRMU2020-17 
Date of Issue 2020-08-26 (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 PRMU2020-17

Conference Information
Committee PRMU  
Conference Date 2020-09-02 - 2020-09-02 
Place (in Japanese) (See Japanese page) 
Place (in English) Online 
Topics (in Japanese) (See Japanese page) 
Topics (in English) Multi-modal, Cross-modal 
Paper Information
Registration To PRMU 
Conference Code 2020-09-PRMU 
Language Japanese 
Title (in Japanese) (See Japanese page) 
Sub Title (in Japanese) (See Japanese page) 
Title (in English) Representation Learning using Video Frame Prediction and Contrastive Learning 
Sub Title (in English)  
Keyword(1) Representation Learning  
Keyword(2) Contrastive Learning  
Keyword(3) Unsupervised Learning  
Keyword(4)  
Keyword(5)  
Keyword(6)  
Keyword(7)  
Keyword(8)  
1st Author's Name Hidemoto Nakada  
1st Author's Affiliation National Institute of Advanced Industrial Science and Technology (AIST)
2nd Author's Name Hideki Asoh  
2nd Author's Affiliation National Institute of Advanced Industrial Science and Technology (AIST)
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 2020-09-02 16:30:00 
Presentation Time 15 minutes 
Registration for PRMU 
Paper # PRMU2020-17 
Volume (vol) vol.120 
Number (no) no.154 
Page pp.59-64 
#Pages
Date of Issue 2020-08-26 (PRMU) 


[Return to Top Page]

[Return to IEICE Web Page]


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