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 2020-09-03 13:10
[Invited Talk] Manifold modeling in embedded space for image restoration
Tatsuya Yokota (Nitech) MI2020-27
Abstract (in Japanese) (See Japanese page) 
(in English) In this invited talk, I will discuss convolutional neural networks, which have achieved remarkable results in various image processing tasks in recent years. First of all, I would like to introduce some studies that are useful for understanding convolutional neural networks, such as Deep Image Prior and Convolutional Sparse Coding. Next, we explain that convolution operation can be separated into delay embedding and linear transformation. From this point of view, we propose to simplify the convolutional neural network to a combination of delay embedding and auto-encoder, and explain that it can be interpreted as a manifold modeling in embedded space (MMES). We show the similarity between Deep Image Prior and MMES by computational experiments, and argue that perspectives such as shift-invariance and self-similarity are useful for understanding the image defined by the structure of the convolutional neural network.
Keyword (in Japanese) (See Japanese page) 
(in English) Convolutional Neural Networks / Deep Image Prior / Delay-embeddin / Auto-encoder / Shift-invariance / Self-similarity / Manifold modeling in embedded space /  
Reference Info. IEICE Tech. Rep., vol. 120, no. 156, MI2020-27, pp. 43-44, Sept. 2020.
Paper # MI2020-27 
Date of Issue 2020-08-27 (MI) 
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 MI2020-27

Conference Information
Committee MI  
Conference Date 2020-09-03 - 2020-09-03 
Place (in Japanese) (See Japanese page) 
Place (in English) Online 
Topics (in Japanese) (See Japanese page) 
Topics (in English) Medical Image Analysis 
Paper Information
Registration To MI 
Conference Code 2020-09-MI 
Language Japanese 
Title (in Japanese) (See Japanese page) 
Sub Title (in Japanese) (See Japanese page) 
Title (in English) Manifold modeling in embedded space for image restoration 
Sub Title (in English)  
Keyword(1) Convolutional Neural Networks  
Keyword(2) Deep Image Prior  
Keyword(3) Delay-embeddin  
Keyword(4) Auto-encoder  
Keyword(5) Shift-invariance  
Keyword(6) Self-similarity  
Keyword(7) Manifold modeling in embedded space  
Keyword(8)  
1st Author's Name Tatsuya Yokota  
1st Author's Affiliation Nagoya Institute of Technology (Nitech)
2nd Author's Name  
2nd Author's Affiliation ()
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 2020-09-03 13:10:00 
Presentation Time 45 minutes 
Registration for MI 
Paper # MI2020-27 
Volume (vol) vol.120 
Number (no) no.156 
Page pp.43-44 
#Pages
Date of Issue 2020-08-27 (MI) 


[Return to Top Page]

[Return to IEICE Web Page]


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