Name
Abstract | ||
|---|---|---|
Given a state-of-the-art deep neural network classifier, we show the existence of a universal (image-agnostic) and very small perturbation vector that causes natural images to be misclassified with high probability. We propose a systematic algorithm for computing universal perturbations, and show that state-of-the-art deep neural networks are highly vulnerable to such perturbations, albeit being quasi-imperceptible to the human eye. We further empirically analyze these universal perturbations and show, in particular, that they generalize very well across neural networks. The surprising existence of universal perturbations reveals important geometric correlations among the high-dimensional decision boundary of classifiers. It further outlines potential security breaches with the existence of single directions in the input space that adversaries can possibly exploit to break a classifier on most natural images. |
Year | DOI | Venue |
|---|---|---|
2017 | 10.1109/CVPR.2017.17 | 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) |
Keywords | DocType | Volume |
universal adversarial perturbations,deep neural network classifier,image-agnostic,very small perturbation vector,natural images,universal perturbations,deep neural networks | Conference | abs/1610.08401 |
Issue | ISSN | ISBN |
1 | 1063-6919 | 978-1-5386-0458-8 |
Citations | PageRank | References |
166 | 5.49 | 16 |
Authors | ||
4 | ||
Name | Order | Citations | PageRank |
|---|---|---|---|
| Seyed-Mohsen Moosavi-Dezfooli | 1 | 627 | 26.32 |
| Alhussein Fawzi | 2 | 766 | 36.80 |
| Omar Fawzi | 3 | 261 | 17.96 |
| Pascal Frossard | 4 | 3015 | 230.41 |