Intrusion detection of multiple attack classes using a deep neural net ensemble

Abstract

An intrusion detection system (IDS) is a necessity to protect against network attacks. The system monitors the activity within a network of connected computers in order to analyze the activity for intrusive patterns. Should an `attack' event happen, then the system has to respond accordingly. Different machine learning techniques have been proposed in the past roughly falling into two categories namely clustering algorithms and classification algorithms. In this paper, the IDS is designed with a neural network ensemble method to classify the different attacks. The neural network ensemble method comprises autoencoder, deep belief neural network, deep neural network, and an extreme learning machine. The NSL-KDD data set is used to measure the detection rate and false alarm rate of the implemented neural network ensemble method. The detection rate and false alarm rate are the two important measure for IDSs, however, several other measures are also reported on such as confusion matrix, classification accuracy, and AUC (area under curve).