On Neural Network approximation of ideal adversarial attack and convergence of adversarial training

AAML
Re-assign community