An Anomaly-based Intrusion Detection System Using Butterfly Optimization Algorithm
Amir Soltany Mahboob, Mohammad Reza Moghaddam · 2020
Intrusion detection system (IDS) is one of the most effective defensive systems which is on high demand due to the increasing number of devices connected to the internet these days. IDS monitors the network packets in order to detect malicious activities. These packets usually have a lot of features most of which are irrelevant and repetitive and can curtail the performance of the IDS. Therefore, it is important to use feature selection techniques to select an optimum subset of features. in this paper, a novel IDS is proposed which employs the butterfly optimization algorithm (BOA), a recent meta-heuristic, to perform feature selection. A multi-layer perceptron (MLP) classifier is used to evaluate the capability of the selected features to predict attacks. In order to improve the MLP classifier, in addition to the gradient descent (GD) training method, two other meta-heuristic methods, particle swarm optimization (PSO) and genetic algorithm (GA) have been used to optimize the classification structure. The NSL-KDD dataset is used in experiments to prove the efficiency of the proposed IDS in detecting four attack types namely: DoS, Probe, R2L and U2R. After repeated tests of the proposed method on four types of attack, the results are reported in terms of classification accuracy. A comparison of classical and intelligent methods in classifier training is made. Comparison with other similar works prove the superiority of the proposed method in reducing the number of features as well as increasing classification accuracy.