Data is growing exponentially fast. It's expected that by the year 2020 1.7 megabytes of data will be generated for each human each second. It's not feasible to label such amount of data and use it for supervised learning. However, it's very tempting to somehow leverage it through unsupervised learning. Generative models are one of the most promising approaches to do that. As the number of parameters of generative models is much lower than amount of data, then those models are forced to discover essential features of data in order to generate it. One may see intuition behind this approach in words of Richard Feynman: "What cannot I create, I do not understand".

In this tutorial we will show how to implement modern generative models in Python such as GANs - Generative Adversarial Networks.