Goal of the project is to define and train a DCGAN on a dataset of faces. The generator network will generate new images of faces that look as realistic as possible. The results of the trained Generator should look like fairly realistic faces with small amounts of noise.