Abstract: Recent advances have heightened the interest in the adversarial transferability of Vision-Language Pre-training (VLP) models. However, most existing strategies constrained by two persistent ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results