A Systematic Survey of Regularization and Normalization in GANs
A Systematic Survey of Regularization and Normalization in GANs
Generative Adversarial Networks (GANs) have been widely applied in different scenarios thanks to the development of deep neural networks. The original GAN was proposed based on the non-parametric assumption of the infinite capacity of networks. However, it is still unknown whether GANs can fit the target distribution without any prior …