arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 1812.04599cs.CVcs.AIcs.LGstat.ML
收藏

用于图像和视频分类的对抗框架

Adversarial Framing for Image and Video Classification

Jagiellonian University(雅盖隆大学);Element AI;Nomagic

Konrad Zolna, Michal Zajac, Negar Rostamzadeh, Pedro O. Pinheiro

摘要:

神经网络容易受到对抗攻击。一般来说,这种攻击通过稍微修改大部分像素或用补丁遮挡像素来降低输入的质量。在本文中,我们提出了一种方法,保持图像不变,只增加了一个对抗框架的图像的边界。我们的经验表明,我们的方法能够成功地攻击最先进的方法对图像和视频分类问题。值得注意的是,所提出的方法导致在测试时非常快的通用攻击。源代码可以在https://github.com/zajaczajac/adv_framing上找到。

英文摘要:

Neural networks are prone to adversarial attacks. In general, such attacks deteriorate the quality of the input by either slightly modifying most of its pixels, or by occluding it with a patch. In this paper, we propose a method that keeps the image unchanged and only adds an adversarial framing on the border of the image. We show empirically that our method is able to successfully attack state-of-the-art methods on both image and video classification problems. Notably, the proposed method results in a universal attack which is very fast at test time. Source code can be found at https://github.com/zajaczajac/adv_framing .

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑