This project follows material outlined in this tutorial, based on a Stanford course with the same title.
main.py
implements a 3-layer autoencoder, which is a fully connected unsupervised neural network.
The data is randomly generated rectangles, source in generate_data.py
.