Two Souls in an Adversarial Image: Towards Universal Adversarial Example Detection using Multi-view Inconsistency

1 code implementation • 25 Sep 2021 • Sohaib Kiani, Sana Awan, Chao Lan, Fengjun Li, Bo Luo

To this end, Argos first amplifies the discrepancies between the visual content of an image and its misclassified label induced by the attack using a set of regeneration mechanisms and then identifies an image as adversarial if the reproduced views deviate to a preset degree.

Adversarial Attack Detection Adversarial Defense +2