Penalty Gradient Normalization for Generative Adversarial Networks

Tian Xia

In this paper, we propose a novel normalization method called penalty gradient normalization (PGN) to tackle the training instability of Generative Adversarial Networks (GANs) caus…