Improving the Intrusion Detection using Discriminative Machine Learning Approach and Improve the Time Complexity by Data Mining Feature Selection Methods

Abstract

As the dependence of daily life is increasing on Internet technology, the attacks on the systems, servers are also rapidly increasing. The motives of attacks are to steal the confidential data from the systems or making the system unavailable to the authorised users. An effective approach is required to detect the intrusions to provide the defence to the Networks. First we applied the feature selection to reduce the dimensions of NSL-KDD data set. By feature reduction and machine learning approach we able to build Intrusion detection model to find attacks on system and improve the intrusion detection using the captured data. The intrusion detection accuracy of learning algorithms is also performed on the data set, without the level 21 attacks which is most easy to identify attacks, using learning algorithms and the success rate of proposed model is calculated over the attacks which are hard to detect.