AI RESEARCH
Unveiling the Backdoor Mechanism Hidden Behind Catastrophic Overfitting in Fast Adversarial Training
arXiv CS.LG
•
ArXi:2604.24350v1 Announce Type: new