arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2002.12527cs.CV
收藏

L2对抗性例子本质上不同吗?

Are L2 adversarial examples intrinsically different?

Brown University(布朗大学);Beihang University(北京航空航天大学)

Mingxuan Li, Jingyuan Wang, Yufan Wu

摘要:

深度神经网络(DDN)在各种任务中取得了显著的成功,包括许多涉及安全的场景。然而,大量的工作已经证明了它对对手的脆弱性。通过理论分析,我们揭示了可以从本质上区分对抗性示例和正常输入的属性。也就是说,由$L_2$攻击生成的对抗性示例通常具有较大的输入敏感度,可以用于有效地识别它们。我们还发现,由$L_\infty$攻击生成的那些将在像素域中有足够的不同以凭经验检测。为了验证我们的分析,我们提出了一个\textbf{G}引导的\textbf{C}补充\textbf{D}防御模块(\textbf{GCD})集成检测和恢复过程。与对抗性检测方法相比,我们的检测器对大多数攻击的检测AUC超过0.98。当将我们的引导矫正器与常用的对抗训练方法和其他矫正方法进行比较时,我们的矫正器的性能大大优于它们。我们在MNIST上实现了高达99\%的恢复分类准确率,在CIFAR-10上实现了89\%,在ImageNet子集上实现了87\%的恢复分类准确率。此外,在白盒设置下,我们的整体防御模块显示出良好的鲁棒性。因此,我们确认,至少$L_2$对抗性的例子是从理论上和经验上的正常输入的本质上是不同的。我们揭示了设计简单而有效的防御方法与这些属性。

英文摘要:

Deep Neural Network (DDN) has achieved notable success in various tasks, including many security concerning scenarios. However, a considerable amount of work has proved its vulnerability to adversaries. We unravel the properties that can intrinsically differentiate adversarial examples and normal inputs through theoretical analysis. That is, adversarial examples generated by $L_2$ attacks usually have larger input sensitivity which can be used to identify them efficiently. We also found that those generated by $L_\infty$ attacks will be different enough in the pixel domain to be detected empirically. To verify our analysis, we proposed a \textbf{G}uided \textbf{C}omplementary \textbf{D}efense module (\textbf{GCD}) integrating detection and recovery processes. When compared with adversarial detection methods, our detector achieves a detection AUC of over 0.98 against most of the attacks. When comparing our guided rectifier with commonly used adversarial training methods and other rectification methods, our rectifier outperforms them by a large margin. We achieve a recovered classification accuracy of up to 99\% on MNIST, 89\% on CIFAR-10, and 87\% on ImageNet subsets against $L_2$ attacks. Furthermore, under the white-box setting, our holistic defensive module shows a promising degree of robustness. Thus, we confirm that at least $L_2$ adversarial examples are intrinsically different enough from normal inputs both theoretically and empirically. And we shed light upon designing simple yet effective defensive methods with these properties.

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。

↑