AI RESEARCH

Unveiling the Backdoor Mechanism Hidden Behind Catastrophic Overfitting in Fast Adversarial Training

arXiv CS.LG

ArXi:2604.24350v1 Announce Type: new