Improving Extreme Learning Machine Accuracy Utilizing Genetic Algorithm for Intrusion Detection Purposes

Abstract

Intrusion detection system (IDS) is a kind of software protection which is developed in order to automatically provide alarms to the supervisor when someone or something tries penetrating the system information by means of malignant activity or by means of security strategy infringements. The extreme learning machine (ELM) is an easy knowledge algorithm for hidden single-layer neural networks SLFNs whose knowledge velocity can be thousands of times quicker than the traditional feeding network learning algorithms such as reverse propagation algorithm (BP) while getting best popularization execution, but the main ELM problem is not more accurate. Genetic algorithms (GAs) have become common as a way in order to provide solutions to the hard combinatorial optimization problems. In this paper, the genetic algorithm will be utilized in order to enhance the ELM accuracy for the IDS purposes. The proposed enhancement is by working and preparing the inputs of the ELM before the processing, and then, the ELM result will be utilized to determine the intrusion.