Adversarial Defense via Data Dependent Activation Function and Total Variation Minimization

Autor:	Wang, Bao, Lin, Alex T., Zhu, Wei, Yin, Penghang, Bertozzi, Andrea L., Osher, Stanley J.
Rok vydání:	2018
Předmět:	Computer Science - Machine Learning Mathematics - Numerical Analysis Statistics - Machine Learning 68Pxx
Zdroj:	Inverse Problems and Imaging, 2020
Druh dokumentu:	Working Paper
Popis:	We improve the robustness of Deep Neural Net (DNN) to adversarial attacks by using an interpolating function as the output activation. This data-dependent activation remarkably improves both the generalization and robustness of DNN. In the CIFAR10 benchmark, we raise the robust accuracy of the adversarially trained ResNet20 from $\sim 46\%$ to $\sim 69\%$ under the state-of-the-art Iterative Fast Gradient Sign Method (IFGSM) based adversarial attack. When we combine this data-dependent activation with total variation minimization on adversarial images and training data augmentation, we achieve an improvement in robust accuracy by 38.9$\%$ for ResNet56 under the strongest IFGSM attack. Furthermore, We provide an intuitive explanation of our defense by analyzing the geometry of the feature space. Comment: 17 pages, 6 figures
Databáze:	arXiv
Externí odkaz:	http://arxiv.org/abs/1809.08516 Zobrazit plný text záznamu View this record from Arxiv