A Deep Q-Learning based approach applied to the Snake game

Abstract

In recent years, one of the highest challenges in the field of artificial intelligence has been the creation of systems capable of learning how to play classic games. This paper presents a Deep Q-Learning based approach for playing the Snake game. All the elements of the related Reinforcement Learning framework are defined. Numerical simulations for both the training and the testing phases are presented. A particular focus is given to the associated Neural Network hyperparameters tuning, which is a crucial step in the agent design process and for the achievement of a desired target level of performance.

Topics & Concepts