Intrusion Detection by SFLA Based Selected Features and Artificial Neural Network

Abstract

 Security of digital nodes in form of network against intrusion is an big issue, as vulnerable network need number of detection measures. This paper has proposed a model where signature based intrusion detection was perform. Proposed model SFLANN is a hybrid combination of genetic algorithm Shuffling Frog Leaching Algorithm SFLA and Error back Propagation neural network. Use of genetic algorithm was done for increasing the detection accuracy by reducing some of input features, as large feature set leads to confusion and it ultimately reduce the accuracy. Use of memeplex concept in SFLA algorithm has increase the work efficiency. Selected features were pass in the EBPNN for training and testing. Real NSL intrusion dataset was used for experiment and comparison. It was obtained that results are better than other existing methods of IDS on precision, recall and accuracy parameters.