Fetching the paper…

Adversarially Robust Training through Structured Gradient Regularization · Around