Unfolding Local Growth Rate Estimates for (Almost) Perfect Adversarial Detection

arXiv:2212.06776v4 Announce Type: replace-cross
Abstract: Convolutional neural networks (CNN) define the state-of-the-art solution on many perceptual tasks. However, current CNN approaches largely remain vulnerable against adversarial perturbations of the input that have been crafted specifically to fool the system while being quasi-imperceptible to the human eye. In recent years, various approaches have been proposed to defend CNNs against such attacks, for example by model hardening or by adding explicit defence mechanisms. Thereby, a small "detector" is included in the network …

adversarial arxiv cs.cr cs.cv detection growth local perfect rate