🤖 AI Summary
The unclear relationship between deep neural network (DNN) uncertainty and adversarial robustness hinders principled robustness assessment. Method: We propose an uncertainty-sensitivity evaluation paradigm integrating Bayesian deep learning, Monte Carlo Dropout, and confidence interval estimation, evaluated across FGSM, PGD, and CW attacks on CIFAR-10/100 and ImageNet subsets. Contribution/Results: We establish the first interpretable mapping between quantitative uncertainty measures—e.g., predictive entropy and confidence—and adversarial robustness. Empirical results show adversarial examples increase average predictive entropy by 47% and reduce confidence by 39%. Critically, abrupt uncertainty shifts enable attack detection up to 0.8 seconds before misclassification. This work provides both a novel theoretical framework for understanding robustness through the lens of epistemic uncertainty and practical tools for proactive adversarial defense and uncertainty-aware evaluation.
📝 Abstract
Deep Neural Networks (DNNs) have achieved state of the art results and even outperformed human accuracy in many challenging tasks, leading to DNNs adoption in a variety of fields including natural language processing, pattern recognition, prediction, and control optimization. However, DNNs are accompanied by uncertainty about their results, causing them to predict an outcome that is either incorrect or outside of a certain level of confidence. These uncertainties stem from model or data constraints, which could be exacerbated by adversarial attacks. Adversarial attacks aim to provide perturbed input to DNNs, causing the DNN to make incorrect predictions or increase model uncertainty. In this review, we explore the relationship between DNN uncertainty and adversarial attacks, emphasizing how adversarial attacks might raise DNN uncertainty.