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 1909.09938cs.LGcs.CR
收藏

HAWKEYE:深度神经网络的对抗性样本检测器

HAWKEYE: Adversarial Example Detector for Deep Neural Networks

Purdue University(普渡大学);Google(谷歌)

Jinkyu Koo, Michael Roth, Saurabh Bagchi

摘要:

对抗性样本(AE)是可以通过向原始图像中引入轻微扰动来误导深度神经网络(DNN)分类器的图像。最近的工作表明,检测AE可能比防止AE产生更有效。然而,最先进的AE检测仍然显示出较高的假阳性率,从而拒绝了相当数量的正常图像。为了解决这个问题,我们提出了HAWKEYE,这是一个单独的神经网络,它分析DNN的输出层,并检测AE。HAWKEYE的AE检测器利用输入图像的量化版本作为参考,并经过训练以区分输入图像上的DNN输出与其参考图像上的DNN输出的变化特性。我们还表明,级联针对不同量化步长训练的AE检测器可以大大降低误报率,同时保持较高的检测率。

英文摘要:

Adversarial examples (AEs) are images that can mislead deep neural network (DNN) classifiers via introducing slight perturbations into original images. Recent work has shown that detecting AEs can be more effective against AEs than preventing them from being generated. However, the state-of-the-art AE detection still shows a high false positive rate, thereby rejecting a considerable amount of normal images. To address this issue, we propose HAWKEYE, which is a separate neural network that analyzes the output layer of the DNN, and detects AEs. HAWKEYE's AE detector utilizes a quantized version of an input image as a reference, and is trained to distinguish the variation characteristics of the DNN output on an input image from the DNN output on its reference image. We also show that cascading our AE detectors that are trained for different quantization step sizes can drastically reduce a false positive rate, while keeping a detection rate high.

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。

↑