PASA: Attack Agnostic Unsupervised Adversarial Detection using Prediction & Attribution Sensitivity Analysis

arXiv:2404.10789v1 Announce Type: new
Abstract: Deep neural networks for classification are vulnerable to adversarial attacks, where small perturbations to input samples lead to incorrect predictions. This susceptibility, combined with the black-box nature of such networks, limits their adoption in critical applications like autonomous driving. Feature-attribution-based explanation methods provide relevance of input features for model predictions on input samples, thus explaining model decisions. However, we observe that both model predictions and feature attributions for input samples are sensitive to noise. We …

adoption adversarial adversarial attacks analysis applications arxiv attack attacks attribution autonomous autonomous driving box classification critical cs.ai cs.cr cs.lg detection driving feature input nature networks neural networks prediction predictions vulnerable