Feature Denoising for Improving Adversarial Robustness

Cihang Xie, Yuxin Wu, Laurens van der Maaten, Alan Yuille, kaiming He, CVPR-2019


This paper proposes a new technique to achieve adversarial robustness for a Deep ConvNet by leveraging the fact that adversarial perturbation lead to noise in the features extracted by these networks. So their network contains noise blocks for denoising feature maps and hence making the network classify the input correctly.

Main contributions

Implementation Details

Our two cents