7 ms·
Deep Learning Adversarial Examples – Clarifying Misconceptions
- SergeyHack 11y agoThe 6th point is very interesting: "Adversarial examples generalize across models trained to perform the same task, even if those models have different architectures and were trained on a different training set."