GAN model architectures

The following section will discuss alternative GAN model architectures for the Discriminator and Generator.

ResNet GAN

In its simplest form, a ResNet network is a network with residual layers. A residual layer is a layer in which the layer input is added to the layer output. This connection from the layer input to the layer output is called a residual connection, as illustrated in the following diagram:

Multiple variations of the a ResNet layer. Copied from Identity Mappings in Deep Residual Networks (https://arxiv.org/abs/1603.05027)

Let's take a look at an implementation of a ResNet GAN that could be used for image denoising, image super resolution, image-to-image translation, audio denoising...