Now showing items 1-1 of 1

    • Analysis of Dominant Classes in Universal Adversarial Perturbations 

      Vadillo, J.; Santana, R.; Lozano, J.A.Autoridad BCAM (2022)
      The reasons why Deep Neural Networks are susceptible to being fooled by adversarial examples remains an open discussion. Indeed, many differ- ent strategies can be employed to efficiently generate adversarial attacks, some ...