1 paper · 1 filter
Xiao Li, Yining Liu, Na Dong +2
Deep learning-based object recognition systems can be easily fooled by various adversarial perturbations. One reason for the weak robustness may be that they do not have part-based…