Research has shown that widely used deep neural networks are vulnerable to carefully crafted adversarial perturbations. Moreover, these adversarial perturbations often transfer across models. We hypothesize that adversarial weakness is composed of three sources of bias: architecture, dataset, and random initialization. We show that one can decompose adversarial examples into an architecture-dependent component, data-dependent component, and noise-dependent component and that these components behave intuitively. For example, noise-dependent components transfer poorly to all other models, while architecture-dependent components transfer better to retrained models with the same architecture. In addition, we demonstrate that these components can be recombined to improve transferability without sacrificing efficacy on the original model.
@article{arxiv.1812.01198,
title = {Adversarial Example Decomposition},
author = {Horace He and Aaron Lou and Qingxuan Jiang and Isay Katsman and Serge Belongie and Ser-Nam Lim},
journal= {arXiv preprint arXiv:1812.01198},
year = {2019}
}
Comments
ICML 2019 Workshop, Security and Privacy of Machine Learning, camera-ready version