| 講演名 | 2020-11-25 GAN based feature-level supportive method for improved adversarial attacks on face recognition Zhengwei Yin(USTC/Hosei Univ.), 内田 薫(法政大), |
|---|---|
| PDFダウンロードページ |  PDFダウンロードページへ |
| 抄録(和) | With the rapid development of deep neural networks (DNN), DNN-based face recognition technologies are also achieving great success and have been widely used in various applications which require high-accuracy and robustness. However, deep neural networks are known to be vulnerable to adversarial attacks, performed using images added with well-designed perturbations. To enhance security of DNN-based face recognition, we need to explore deeper the mechanisms of related technologies. In this paper, we propose a feature-level supportive method, BiasGAN, to improve the performance of universal adversarial attack methods. We insert this image to image translation preprocessor before conducting adversarial example generation. BiasGAN will search in the potential face space and can generate images with biased face feature, causing generated face images to be easier to perturb efficiently. Experimental results show that this approach improves both fooling ratio and average perturbation size significantly at different perturbation levels. |
| 抄録(英) | With the rapid development of deep neural networks (DNN), DNN-based face recognition technologies are also achieving great success and have been widely used in various applications which require high-accuracy and robustness. However, deep neural networks are known to be vulnerable to adversarial attacks, performed using images added with well-designed perturbations. To enhance security of DNN-based face recognition, we need to explore deeper the mechanisms of related technologies. In this paper, we propose a feature-level supportive method, BiasGAN, to improve the performance of universal adversarial attack methods. We insert this image to image translation preprocessor before conducting adversarial example generation. BiasGAN will search in the potential face space and can generate images with biased face feature, causing generated face images to be easier to perturb efficiently. Experimental results show that this approach improves both fooling ratio and average perturbation size significantly at different perturbation levels. |
| キーワード(和) | Deep neural network / enerative adversarial network / Face recognition / Adversarial attack |
| キーワード(英) | Deep neural network / enerative adversarial network / Face recognition / Adversarial attack |
| 資料番号 | BioX2020-35 |
| 発行日 | 2020-11-18 (BioX) |
| 研究会情報 | |
| 研究会 | BioX |
|---|---|
| 開催期間 | 2020/11/25(から1日開催) |
| 開催地（和） | オンライン開催 |
| 開催地（英） | Online |
| テーマ（和） | バイオメトリクス全般 (International session含む） |
| テーマ（英） | |
| 委員長氏名（和） | 大塚 玲(産総研) |
| 委員長氏名（英） | Akira Otsuka(AIST) |
| 副委員長氏名（和） | 青木 隆浩(富士通研) / 市野 将嗣(電通大) |
| 副委員長氏名（英） | Takahiro Aoki(Fujitsu Labs.) / Masatsugu Ichino(Univ. of Electro-Comm.) |
| 幹事氏名（和） | 高田 直幸(セコム) / 奥井 宣広(KDDI総合研究所) |
| 幹事氏名（英） | Naoyuki Takada(SECOM) / Norihiko Okui(KDDI Research) |
| 幹事補佐氏名（和） | 佐野 恵美子(三菱電機) / 早坂 昭裕(NEC) |
| 幹事補佐氏名（英） | Emiko Sano(MitsubishiElectric) / Akihiro Hayasaka(NEC) |
| 講演論文情報詳細 | |
| 申込み研究会 | Technical Committee on Biometrics |
|---|---|
| 本文の言語 | ENG |
| タイトル（和） | |
| サブタイトル（和） | |
| タイトル（英） | GAN based feature-level supportive method for improved adversarial attacks on face recognition |
| サブタイトル（和） | |
| キーワード(1)（和/英） | Deep neural network / Deep neural network |
| キーワード(2)（和/英） | enerative adversarial network / enerative adversarial network |
| キーワード(3)（和/英） | Face recognition / Face recognition |
| キーワード(4)（和/英） | Adversarial attack / Adversarial attack |
| 第 1 著者 氏名（和/英） | Zhengwei Yin / Zhengwei Yin |
| 第 1 著者 所属（和/英） | University of Science and Technology of China/Hosei University(略称：USTC/Hosei Univ.) University of Science and Technology of China/Hosei University(略称：USTC/Hosei Univ.) |
| 第 2 著者 氏名（和/英） | 内田 薫 / Kaoru Uchida |
| 第 2 著者 所属（和/英） | 法政大学(略称：法政大) Hosei University(略称：Hosei Univ.) |
| 発表年月日 | 2020-11-25 |
| 資料番号 | BioX2020-35 |
| 巻番号（vol） | vol.120 |
| 号番号（no） | BioX-247 |
| ページ範囲 | pp.1-6(BioX), |
| ページ数 | 6 |
| 発行日 | 2020-11-18 (BioX) |