In Chapter 9, we delved into the generative and creative side of deep learning. There, we used two recurrent neural networks to forecast and generate classical-looking passages of text. But text data, or generally speaking, sequential data, is not the only thing that these neural networks can create. Images are possible too.
If there is a recent and radical algorithm that has taken over the machine learning field, it has to be the generative adversarial networks (Goodfellow et al., 2014), or GANs. GANs—which ...
