DLAI Project Team 6

This is the project repository for DLAI Team 6. The Team is made up by:

Alvaro Borreguero Arnau Cusco Ignacio Penas Rubén Díez
Álvaro Borreguero Arnau Cusco Ignacio Penas Rubén Díez

This project was developed for the Telecombcn-Deep Learning for Artificial Intelligence course at UPC during Fall Semester 2017.

Universitat Polit�cnica de Catalunya

Introduction

The main objective of the project is to optimize the execution of the game Ms. Pac-Man. We lack experience in either Machine Learning or Deep Learning, therefore we decided to take as project, the optimization of an existant project proposal. The project introduced reinforcement learning as a major innovation and objective. In order to determine which parameters are going to optimize the program, we will propose some changes that we believe will make a significant difference. This may lead us to some different results in terms of accuracy and loss, so we can conclude which metric and values are the most accurate.