In this chapter, we discussed how to create new images with generative models, which is one of the most exciting machine learning areas at the moment. We talked about two of the most popular generative algorithms: VAEs and GANs. First, we learned their theoretical foundations and then we implemented simple programs to generate new MNIST digits with each algorithm.
This chapter concludes the series of the last three chapters, which were dedicated to computer vision. In the next chapter, we'll discuss how to apply DL algorithms in the field of natural language processing (NLP). We'll also introduce the main NLP paradigms and a new type of neural network, called the recurrent network, which is especially suited for NLP tasks.