The concept of style transfer refers to the process of rendering the style of a product into another product. Imagine that your fashion-crazy friend bought a blue-printed bag and wanted to get a pair of shoes of a similar print to go with it. Up until 2016, this might not have been possible, unless they were friends with a fashion designer who would first have to design a shoe before it was approved for production. With the recent progress in generative adversarial networks, however, this kind of design process can be carried out easily.

A generative adversarial network is a network that learns by playing a zero sum game between a generator network and a discriminator network. Let's say that a fashion designer wants to design a handbag of a specific structure and is exploring different prints. The designer might sketch the structure...