Universal Adversarial Perturbations: Efficiency on a small image dataset

Waris Radji

Although neural networks perform very well on the image classification task, they are still vulnerable to adversarial perturbations that can fool a neural network without visibly c…