| 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 | 2023-01-26 09:55 On the Transferability of Adversarial Examples between Isotropic Network and CNN models Miki Tanaka (Tokyo Metropolitan Univ.), Isao Echizen (NII), Hitoshi Kiya (Tokyo Metropolitan Univ.)  EMM2022-62 |
|
| Abstract | (in Japanese) | (See Japanese page) |
| (in English) | Deep neural networks are well known to be vulnerable to adversarial examples (AEs). In addition, AEs generated for a source model fool other (target) models, and this property is called adversarial transferability. In this paper, we investigate the transferability between a convolutional neural network (CNN) as ResNet and an Isootropic network such the vision transformer (ViT) and ConvMixer. In addition, the use of encrypted models is evaluated in terms of the transferability of AEs. In an experiment, the transferability of ViT was confirmed to be low. Furthermore, the use of encrypted models was confirmed to reduce the influence of the transferability between models. | |
| Keyword | (in Japanese) | (See Japanese page) |
| (in English) | Adversarial example / Transferability / Deep learning / / / / / | |
| Reference Info. | IEICE Tech. Rep., vol. 122, no. 368, EMM2022-62, pp. 7-12, Jan. 2023. | |
| Paper # | EMM2022-62 | |
| Date of Issue | 2023-01-19 (EMM) | |
| 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 | EMM2022-62 |
|
| Conference Information | |
| Committee | EMM |
| Conference Date | 2023-01-26 - 2023-01-26 |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | Tohoku Univ. |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | Sense of Presence, Universal Media, Digital Entertainment, etc. |
| Paper Information | |
| Registration To | EMM |
| Conference Code | 2023-01-EMM |
| Language | Japanese |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | On the Transferability of Adversarial Examples between Isotropic Network and CNN models |
| Sub Title (in English) | |
| Keyword(1) | Adversarial example |
| Keyword(2) | Transferability |
| Keyword(3) | Deep learning |
| Keyword(4) | |
| Keyword(5) | |
| Keyword(6) | |
| Keyword(7) | |
| Keyword(8) | |
| 1st Author's Name | Miki Tanaka |
| 1st Author's Affiliation | Tokyo Metropolitan University (Tokyo Metropolitan Univ.) |
| 2nd Author's Name | Isao Echizen |
| 2nd Author's Affiliation | National Institute of Informatics (NII) |
| 3rd Author's Name | Hitoshi Kiya |
| 3rd Author's Affiliation | Tokyo Metropolitan University (Tokyo Metropolitan 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 | () |
| Speaker | Author-1 |
| Date Time | 2023-01-26 09:55:00 |
| Presentation Time | 25 minutes |
| Registration for | EMM |
| Paper # | EMM2022-62 |
| Volume (vol) | vol.122 |
| Number (no) | no.368 |
| Page | pp.7-12 |
| #Pages | 6 |
| Date of Issue | 2023-01-19 (EMM) |
[Return to Top Page]