Fetching the paper…

Improving Transferability of Adversarial Examples with Input Diversity · Around