COMPARISON OF HYBRID MACHINE LEARNING ALGORITHMS FOR INTRUDER DETECTION

1M. Chithik Raja, M. Munir Ahmed Rabbani

111 Views
40 Downloads
Abstract:

In the modern world Network information security has become an important concern of Internet and volume data. Web security services 'intrusion detection system (IDS) is a critical component that uses network traffic statistics to identify attacks. IDS should be able to implement data exploration and information mining systems to classify network machine outbreaks. The computation costs of IDS, however, is also necessary to help with the actual detection in line for the dismissal and inappropriate features in the Internet stream of traffic-dataset. We in this manner break down six Artificial Intelligence algorithm for IDS. Everyplace we independently execute information preprocessing with two sorts of dimensionality decrease procedures like Principal Component Analysis and Singular Value Decomposition to detect the attacks in the NSL-KDD dataset. The investigation results on the NSL-KDD dataset check that the course of action estimations with dimensionality drop out stands in discovery rate and speed. From our experimental result Principle Component Analysis with classifier KNN gives its matchless quality comparing with Singular Value Decomposition technique.

Keywords:

IDS, SVM, KNN, NSL-KDD

Paper Details
Month2
Year2020
Volume24
IssueIssue 2
Pages5431-5439