Stabilizing Training of Generative Adversarial Networks through Regularization