This repository is a demonstration on how to generate human faces using Generative Adversarial Network (GAN). There are many different GAN models out there that are built for specific data generation task. This GAN implementation is adapted from the one featured in HSU's Introduction to Deep Learning course: https://github.com/hse-aml/intro-to-dl/blob/master/week4/Adversarial-task.ipynb
In addition, I owe much of the code written to download and prepare the data files and training the model from https://github.com/dimitry-ishenko-ml/. Thank you very much.
The model was trained for 30000 epochs, each with 100 images. The results came out pretty good and the faces are relatively realistic. Feel free to train for much longer and observe to see if GAN can produce better images.