Colored Noise Injection for Training Adversarially Robust Neural Networks
Machine Learning
2020-03-23 v2 Computer Vision and Pattern Recognition
Machine Learning
Abstract
Even though deep learning has shown unmatched performance on various tasks, neural networks have been shown to be vulnerable to small adversarial perturbations of the input that lead to significant performance degradation. In this work we extend the idea of adding white Gaussian noise to the network weights and activations during adversarial training (PNI) to the injection of colored noise for defense against common white-box and black-box attacks. We show that our approach outperforms PNI and various previous approaches in terms of adversarial accuracy on CIFAR-10 and CIFAR-100 datasets. In addition, we provide an extensive ablation study of the proposed method justifying the chosen configurations.
Cite
@article{arxiv.2003.02188,
title = {Colored Noise Injection for Training Adversarially Robust Neural Networks},
author = {Evgenii Zheltonozhskii and Chaim Baskin and Yaniv Nemcovsky and Brian Chmiel and Avi Mendelson and Alex M. Bronstein},
journal= {arXiv preprint arXiv:2003.02188},
year = {2020}
}