Adversarial Training of Two-Layer Polynomial and ReLU Activation Networks via Convex Optimization

Open in new window