Neural networks algorithm as an advisory tool in cyberattribution

Abstract

Cyber ​​attribution is a fundamental part of a State's cyber defense. The task of assigning someone responsible for a cyber-aggression (and especially if it is another State) is really complicated, taking into account the technological advancement of tools related to the objectives of cyber-attackers. This activity (that is, Cyberattribution) is essential to create a true deterrent that discourages the execution of the aforementioned attacks. On the other hand, computer science has rapidly developed the theory and use of hundreds of artificial intelligence tools that are seen in our daily lives, and in which neural networks are used. Neural networks are fed by thousands of data and their results are more than acceptable for optimization, classification or prediction. The sources of entry of these data can be totally varied and according to their quantity it can be affirmed that it will be possible to achieve greater or lesser precision in the output. It is proposed to demonstrate that neural networks in the context of Cyberattribution procedures can be used successfully as a tool for advising on the determination of the origin of a cyber attack on an infrastructure that has a critical function. Although there are studies referring to this subject, the particularity of this thesis is that of registering in the field of cyber defense itself in a whole in accordance with current national legislation.