Machine learning for Smartphone games

This thesis concerns the creation of an artificial player for a smartphone game able to learn the rules and how to win by itself thanks to Machine Learning techniques. The chosen game is an adaptation of the well known "rock-paper-scissors" game. The first step of this work is to clearly define the game with frameworks game theory and Machine Learning and to analyze the imposed constraint which is the smartphone hardware. The second step is to create the agent according to these frameworks by improving and assessing its performances. This part will also include the final results obtained and show that it can compete against a human.