Practical Convex Formulations of One-hidden-layer Neural Network Adversarial Training | IEEE Conference Publication | IEEE Xplore