Study of the Attributes using Four Class Labels on KDD99 and NSL-KDD Datasets with Machine Learning Techniques

Abstract

An effective dataset results in a smart intrusion inspection system. The paramount target of this field of examination is to acculturate about the contribution of attributes in KDD99 and NSL-KDD datasets. In this paper, divergent machine learning algorithms are employed with respect to four classes of attacks and their performances are compared for detecting the abnormalities present in the network traffic paradigms. The relationship between networks protocols with the attacks is also examined using data reduction techniques as an input to the selected classification algorithms for observing the performances of the algorithms on both the datasets. The study unfolds various facts about the network attacks and the protocols.