Training Guidance with KDD Cup 1999 and NSL-KDD Data Sets of ANIDINR: Anomaly-Based Network Intrusion Detection System
Benedetto Marco Serinelli, Anastasija Collen, Niels Alexander Nijdam
Abstract
In today's world, the protection of the computer networks remains one of the most crucial and difficult challenges in cyber security. In this work, a passive defence system ANIDINR is presented, aiming to monitor and protect computer networks. Our effort is focused on providing step-by-step guidance on methodologies selection and execution for the Machine and Deep Learning models' training. Taking as an input two data sets, five MDL models are evaluated. Our goals are to minimise the percentage of Undetected Attack, the percentage of False Alarm Rate and the overall testing time. Based on this set-up, the proposed system is capable to predict in near-to-real time well-known and zero-day computer network attacks.