1 paper · 1 filter
Xiaosen Wang, Zhijin Ge, Bohan Liu +5
Adversarial transferability refers to the capacity of adversarial examples generated on the surrogate model to deceive alternate, unexposed victim models. This property eliminates…