Regularizing Neural Networks via Adversarial Model Perturbation | IEEE Conference Publication | IEEE Xplore