1 paper · 1 filter
Tao Sun, Lu Pang, Weimin Lyu +2
Deep neural networks are vulnerable to backdoor attacks, where an adversary manipulates the model behavior through overlaying images with special triggers. Existing backdoor defens…