Analysis, Design, and Comparison of Machine-Learning Techniques for Networking Intrusion Detection

The use of machine-learning techniques is becoming more and frequent in solving all those problems where it difficult to rationally interpret the process interest. Intrusion detection networked systems a problem which, although not fundamental measures that one able obtain from process, important an answer classification algorithm if network traffic characterized by anomalies (and hence, there high probability intrusion) or not. Due increased adoption SW-defined autonomous are distributed interconnected, cyber attack increased, as well its consequence terms system reliability, availability, even safety. In this work, we present application different models anomaly context local area (LAN) analysis. particular, K-nearest neighbors (KNN) artificial neural (ANN) realize for intrusion (IDS). dataset used work representative communication common LAN networks military particular typical US Air Force LAN. This presents training phase based on multidimensional-scaling preprocessing procedure, metrics, provide higher performance generalization with respect model prediction capability. obtained results KNN ANN classifiers compared commonly index evaluation.