Cyber-attribution is a fundamental part of the cyber-defense of a State because it allows to deduce the origin of a cyber-attack on its critical infrastructures. Given that these actions try to hide their origin, the task of assigning a responsible is really complicated considering the technological advance and the vertiginous development of tools related to the objectives of the cyberattackers. On the other hand, computer science has developed in this century in an accelerated way the theory and use of hundreds of artificial intelligence tools. This thesis aims to demonstrate that neural networks in the context of cyber-attribution procedures can be successfully employed as a tool to assist in determining the origin of a cyber-attack.