Adversarial Training on Purification (AToP): Advancing Both Robustness and Generalization.

Guang Lin, Chao Li, Jianhai Zhang, Toshihisa Tanaka, Qibin Zhao

Browse the full ICLR paper archive.