The goal of this project is to train a simple neural network to classify numbers from the MNIST dataset. Then, the weight values are ported to Minecraft, where they are used to natively classify numbers drawn by the user.

The code is split into two sections---model and mc--responsible for the training of the model in python and the porting of the model into mcfunction respectively. Note, the generate_test.py file is incorrectly placed in the model directory because I don't know how to access the preprocessing file from the mc directory (slight skill issue).

Original code for preprocessing.py and visualize.py from Brown University course CSCI1470 HW2. Original code from main.py test and train functions from Brown University course CSCI1470 HW5.

Heavily modified by Taj Gillin for use in DeepMine.