For the machine learning track of the University of Idaho's inaugural Data Science competition, students were tasked with constructing a neural network that could grade handwritten arithmetic problems. These handwritten problems took one of four possible forms:
a + b = c,a - b = c,a = b + c, ora = b - c.
This repository contains the scripts which trained the first place network.
This README will eventually contain a summary of my experiences as a participant in this competition, as well as a thorough description of the network design.