Universal adversarial perturbations against object detection
文献类型：期刊论文
| 作者 | Li, Debang1,3 ; Zhang, Junge1,3; Huang, Kaiqi1,2,3
|
| 刊名 | PATTERN RECOGNITION
 |
| 出版日期 | 2021-02-01 |
| 卷号 | 110页码:12 |
| 关键词 | Adversarial examples Object detection Universal adversarial perturbation |
| ISSN号 | 0031-3203 |
| DOI | 10.1016/j.patcog.2020.107584 |
| 通讯作者 | Huang, Kaiqi(kqhuang@nlpr.ia.ac.cn) |
| 英文摘要 | Despite the remarkable success of deep neural networks on many visual tasks, they have been proved to be vulnerable to adversarial examples. For visual tasks, adversarial examples are images added with visu-ally imperceptible perturbations that result in failure for recognition. Previous works have demonstrated that adversarial perturbations can cause neural networks to fail on object detection. But these methods focus on generating an adversarial perturbation for a specific image, which is the image-specific perturbation. This paper tries to extend such image-level adversarial perturbations to detector-level, which are universal (image-agnostic) adversarial perturbations. Motivated by this, we propose a Universal Dense Object Suppression (U-DOS) algorithm to derive the universal adversarial perturbations against object detection and show that such perturbations with visual imperceptibility can lead the state-of-the-art detectors to fail in finding any objects in most images. Compared to image-specific perturbations, the results of image-agnostic perturbations are more interesting and also pose more challenges in AI security, because they are more convenient to be applied in the real physical world. We also analyze the generalization of such universal adversarial perturbations across different detectors and datasets under the black-box attack settings, showing it's a simple but promising adversarial attack approach against object detection. Furthermore, we validate the class-specific universal perturbations, which can remove the detection results of the target class and keep others unchanged. (c) 2020 Published by Elsevier Ltd. |
| 资助项目 | National Natural Science Foundation of China[61876181] ; National Natural Science Foundation of China[61673375] ; National Natural Science Foundation of China[61721004] ; Projects of Chinese Academy of Sciences[QYZDB-SSW-JSC006] |
| WOS研究方向 | Computer Science ; Engineering |
| 语种 | 英语 |
| WOS记录号 | WOS:000585302200003 |
| 出版者 | ELSEVIER SCI LTD |
| 资助机构 | National Natural Science Foundation of China ; Projects of Chinese Academy of Sciences |
| 源URL | [http://ir.ia.ac.cn/handle/173211/41677]  |
| 专题 | 智能系统与工程 |
| 通讯作者 | Huang, Kaiqi |
| 作者单位 | 1.Univ Chinese Acad Sci, Sch Artificial Intelligence, Beijing, Peoples R China 2.CAS Ctr Excellence Brain Sci & Intelligence Techn, Beijing, Peoples R China 3.Chinese Acad Sci, Inst Automat, CRISE, Beijing, Peoples R China |
| 推荐引用方式 GB/T 7714 | Li, Debang,Zhang, Junge,Huang, Kaiqi. Universal adversarial perturbations against object detection[J]. PATTERN RECOGNITION,2021,110:12. |
| APA | Li, Debang,Zhang, Junge,&Huang, Kaiqi.(2021).Universal adversarial perturbations against object detection.PATTERN RECOGNITION,110,12. |
| MLA | Li, Debang,et al."Universal adversarial perturbations against object detection".PATTERN RECOGNITION 110(2021):12. |
入库方式： OAI收割
来源：自动化研究所
浏览0
下载0
收藏0
其他版本
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。