We propose a novel data-free universal adversarial attack method that leverages semantic samples drawn from a pseudo-semantic prior. The implementation is based on PyTorch. The commands below initiate ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results