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 2022-03-07 16:35
Access Control of Vision Transformer models without performance degradation
Ryota Iijima, Yuma Kinoshita, Hitoshi Kiya (Tokyo Metro. Univ.) EMM2021-104
Abstract (in Japanese) (See Japanese page) 
(in English) In this paper, we propose a novel access control method to protect trained models from unauthorized access. Conventional access control methods cause some performance degradation, and need additional calculations to offer an access control function to models in general. In addition, models have to be trained with specially prepared data such as encrypted images. In contrast, the proposed method does not require any specially prepared data and network modification, and does not cause any performance degradation. The proposed method focuses on the patching process of the vision transformer (ViT) and the structure of the position embedding. We point out that the trained model can be directly encrypted on the bases of the architecture of ViT, and the encrypted models are effective in access control. In an experiment, the effectiveness of the proposed method is evaluated in terms of performance degradation and access control in an image classification task under the use of the CIFAR-10 dataset.
Keyword (in Japanese) (See Japanese page) 
(in English) Vision Transformer / access control / / / / / /  
Reference Info. IEICE Tech. Rep., vol. 121, no. 417, EMM2021-104, pp. 70-75, March 2022.
Paper # EMM2021-104 
Date of Issue 2022-02-28 (EMM) 
ISSN 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 EMM2021-104

Conference Information
Committee EMM  
Conference Date 2022-03-07 - 2022-03-08 
Place (in Japanese) (See Japanese page) 
Place (in English) (Primary: Online, Secondary: On-site) 
Topics (in Japanese) (See Japanese page) 
Topics (in English) Image and Sound Quality, Metrics for Perception and Recognition, Human Auditory and Visual System, etc. 
Paper Information
Registration To EMM 
Conference Code 2022-03-EMM 
Language Japanese 
Title (in Japanese) (See Japanese page) 
Sub Title (in Japanese) (See Japanese page) 
Title (in English) Access Control of Vision Transformer models without performance degradation 
Sub Title (in English)  
Keyword(1) Vision Transformer  
Keyword(2) access control  
1st Author's Name Ryota Iijima  
1st Author's Affiliation Tokyo Metropolitan University (Tokyo Metro. Univ.)
2nd Author's Name Yuma Kinoshita  
2nd Author's Affiliation Tokyo Metropolitan University (Tokyo Metro. Univ.)
3rd Author's Name Hitoshi Kiya  
3rd Author's Affiliation Tokyo Metropolitan University (Tokyo Metro. 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 2022-03-07 16:35:00 
Presentation Time 25 minutes 
Registration for EMM 
Paper # EMM2021-104 
Volume (vol) vol.121 
Number (no) no.417 
Page pp.70-75 
Date of Issue 2022-02-28 (EMM) 

[Return to Top Page]

[Return to IEICE Web Page]

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