Anomaly Detection Using Feed Forward and Radial Basis Function Neural Network

This paper presents the design and development of intelligent system for anomaly detection using neural networks. This paper presents the application of Multi Layer Feed Forward (MLFF) neural network approach and Radial Basis Function (RBF) neural network for anomaly detection. The KDD Cup 1999 dataset is considered as a benchmark dataset for evaluating this algorithm. The network is developed to detect a total of twenty three, in which twenty two intruders and a normal. The training and testing data required to develop the model were randomly selected subset that contain both attacks and normal records from KDD Cup 1999 dataset. Principal Component Analysis based feature extraction method is used for dimensionality reduction. An integration platform used for developing this intelligent system is Visual Basic and Mat lab. The performance of the anomaly detection system using RBF is found to be best when compared to the MLFF results. The comparison of results and consistency are presented in this paper.