Novel Machine Learning Technique for Intrusion Detection in Recent Network-based Attacks

Intrusion Detection is a vastly growing area. Traditionally supervised learning techniques were used for detecting intrusions in the network traffic data. But nowadays not only the rate of traffic has increased tremendously, but the nature of network attacks is also changing. Detecting these new types of attacks requires improvements in detection techniques. Machine learning algorithms are well researched by researchers for detecting anomalies in the network traffic. New datasets have been added in the public repositories. In this paper, we have used novel feature reduction based machine learning algorithms for detecting anomalous patterns in the recently provided dataset. High accuracy of 86.15 percent has been achieved.