Aplicacion de las Redes Neuronales al Reconocimiento de Sistemas Operativos

TL;DR

This paper applies multi-layer perceptron neural networks to the problem of remote operating system identification, demonstrating improved classification results over traditional methods in information security contexts.

Contribution

It introduces the use of neural networks for OS recognition, providing detailed algorithms and showing superior performance compared to classic techniques.

Findings

Neural networks outperform traditional classification methods in OS detection.

Detailed algorithms for training multi-layer perceptrons are provided.

Effective application of AI techniques to information security problems.

Abstract

In this work we present a family of neural networks, the multi-layer perceptron networks, and some of the algorithms used to train those networks (we hope that with enough details and precision as to satisfy a mathematical public). Then we study how to use those networks to solve a problem that arises from the field of information security: the remote identification of Operating Systems (part of the information gathering steps of the penetration testing methodology). This is the contribution of this work: it is an application of classic Artificial Intelligence techniques to a classification problem that gave better results than the classic techniques used to solve it.