Machine learning techniques applied to intruder detection in networks

The intrusion in data networks, are a constant problem faced by networks administrators. Because of this, it is necessary identify, study and propose techniques to detect the moment in which the network is attacked, with the purpose of take measures to mitigate these threats. In this paper was conducted a study of the threats taxonomy that could lead to an attack in a data network. For this, we have identified the most relevant characteristics of the network traffic in order to be processed and classified using machine learning techniques, specifically the normalization (Z-Score), dimensionality reduction (PCA) and classification based on artificial neural networks (ANN) to suggest an intrusion detection system (IDS).