An Empirical Study of Aegis

arXiv:2404.15784v1 Announce Type: new
Abstract: Bit flipping attacks are one class of attacks on neural networks with numerous defense mechanisms invented to mitigate its potency. Due to the importance of ensuring the robustness of these defense mechanisms, we perform an empirical study on the Aegis framework. We evaluate the baseline mechanisms of Aegis on low-entropy data (MNIST), and we evaluate a pre-trained model with the mechanisms fine-tuned on MNIST. We also compare the use of data augmentation to the robustness …

abstract arxiv attacks class cs.lg defense entropy framework importance low networks neural networks robustness study type