Abstract: In this paper, we present a novel task of source-free cross-modal adversarial example generation, which generates adversarial examples based on textual descriptions of attackers. This task ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results