Convergence and Margin of Adversarial Training on Separable Data

AAML
Re-assign community