Public domain datasets for optimizing network intrusion and machine learning approaches

Abstract

Network intrusion detection system (NIDS) commonly attributed to the task to mitigate network and security attacks that has potential to compromise the safety of a network resources and its information. Research in this area mainly focuses to improve the detection method in network traffic flow. Machine learning techniques had been widely used to analyze large datasets including network traffic. In order to develop a sound mechanism for NIDS detection tool, benchmark datasets is required to assist the data mining process. This paper presents the benchmark datasets available publicly for NIDS study such as KDDCup99, IES, pcapr and others. We use some popular machine learning tools to visualize the properties and characteristics of the benchmark datasets.