In the first following section we will discuss the process of generating random variables from a given distribution. Then, in section 2 we will show, through an example, that the problems GANs try to tackle can be expressed as random variable generation problems. In section 3 we will discuss matching based generative networks and show how they answer problems described in section 2. Finally in section 4 we will introduce GANs. More especially, we will present the general architecture with its loss function and we will make the link with all the previous parts.
Don't forget to tag @Roccabat in your comment, otherwise they may not be notified.