Continual Adversarial Defense: Overcoming Catastrophic Forgetting in Defending Against Evolving Attacks
Continual adversarial defense is crucial for ensuring the reliability of deep neural networks in real-world deployment scenarios, where new attacks can emerge in sequences. The authors propose a lifelong defense baseline called Anisotropic & Isotropic Replay (AIR) to alleviate catastrophic forgetting when adapting to new attacks.