Artificial Neural Network based Intrusion Detection System: A Survey

Abstract

Detecting unknown or modified attacks is one of the recent challenges in the field of IDS. Anomaly based IDS can play a very important role in this case. In the first part of this paper, we will focus on how ANN is recently used to address these issues. Number of the researchers has already shown the importance of the various Artificial Neural Network (ANN) based techniques for anomaly detection. In this paper, we will focus on Simple and Hybrid ANN based approach for anomaly detection. In simple approach we will discuss on how Back Propagation Neural Network (BPNN), Self Organizing Maps (SOM), Support Vector Machine (SVM), and Simulated Annealing Neural Network (SA) are used for anomaly detection? While in hybrid approach, we will focus on how more than one above technique are used? In the second part of the paper, we will try to compare the different ANN based techniques in terms of training time, number of the epochs required, converge rate, detection rate, learning approach, etc. Finally we will provide guidelines for the future work.