Advanced Deep Learning with Python

Wasserstein GAN

To understand the Wasserstein GAN (WGAN, https://arxiv.org/abs/1701.07875), let's recall that, in the Training GANs section, we denoted the probability distribution of the generator with and the probability distribution of the real data with . In the process of training the GAN model, we update the generator weights and so we change . The goal of the GAN framework is to converge to (this is also valid for other types of generative ...