Introduction

Introduction

The framework consists of several components: a simulation of the game Ms. Pac-Man, several input algorithms which can numerically represent the in-game situation, one or more neural networks to perform the decision making, static rewards based on game performance, reinforcement learning algorithms and two learning rules to train the neural networks.

To find out how these components were set-up and how they work together, read one of the publications. They provide important details on the various aspects of the framework.
All components can be configured and tweaked using the instructions found in [Configuration].