PacMan Neural Network

Purpose: To develop a neural network that learns to navigate and succeed in a PacMan inspired game.

Background:

I build this project to further my experience with Python, but more specifically PyTorch and reinforcement learning. I had an idea to create a real-estate machine learning program, but no experience with machine learning, so this project's real purpose was as a hands-on and application based teaching tool for myself. This project uses a Q-Learning algorithmn to develop it's decision making, and it is rewarded and punished for moves that I deemed productive or counter productive.