Anomaly Detection in Network using Genetic Algorithm and Support Vector Machine

Abstract- An anomaly is a abnormal activity or deviation from the normal behaviour.Anomaly detection is the process of removing these abnormal or anomalous behaviour from data or services. In this paper, we present a novel method for the detection of anomaly in network. The proposed detection algorithm, is a hybrid algorithm. It is combination of two algorithm genetic and SVM. Experimental results demonstrate to be superior than existing k-mean algorithm. One of the most common problems in existing K means detection techniques is that one must specify the clusters in advance and further the algorithm is very sensitive of noise, mixed pixels and outliers. The definition of means limit the application to only numerical variables. It is data driven with relatively few assumptions on the distributions of underlying data. This paper investigates the performances of genetic algorithm (GA) with support vector machine (SVM) classification method for detecting different types of network attacks.. Genetic based feature selection algorithm reduces the 41 features of the KDD cup dataset into 9 important features by applying fitness value as a threshold and then these 9 features are used for classification using support vector machine. In this work GA and SVM have been implemented and tested on KDD CUP 1999 dataset. Our method has more accurate as compare to existing once. Keywords-Anomaly detection techniques, clustering, CAD, genetic and classification based technique. 1