2 Comments
Oct 22, 2020Liked by Ksenia Se

The easiest way to generate adversarial examples remains gradient-based attacks, by reading the first part we could think that GANs are almost the only way to go.

Expand full comment