Adversarial Feature Alignment: Balancing Robustness and Accuracy in Deep Learning via Adversarial Training

arXiv:2402.12187v1 Announce Type: cross
Abstract: Deep learning models continue to advance in accuracy, yet they remain vulnerable to adversarial attacks, which often lead to the misclassification of adversarial examples. Adversarial training is used to mitigate this problem by increasing robustness against these attacks. However, this approach typically reduces a model's standard accuracy on clean, non-adversarial samples. The necessity for deep learning models to balance both robustness and accuracy for security is obvious, but achieving this balance remains challenging, and the …

accuracy adversarial adversarial attacks alignment arxiv attacks continue cs.cr cs.cv cs.lg deep learning examples feature problem robustness training vulnerable