Pattern based network security using decision trees and support vector machine

Decision tree is an important method for both induction research and data mining, which is mainly used for model classification and prediction. WEKA is software which is capable of doing work on various decision tree algorithms and support vector machine. In this paper, the comparative study of all Decision Tree algorithms is done. The training time, Accuracy and size of tree are the parameters used as performance measures. It is concluded that J48 Graft algorithm performs better than other algorithms. The dataset used is KDD cup´99 dataset. This dataset contains normal as well as abnormal packets. The dataset is highly uneven. We worked out on some 1000 selected packets. The support vector machine algorithms have the ability to be trained and `learn´ in a given environment. This feature can be used in connection with an intrusion detection system, where the support vector machine algorithm can be trained to detect intrusions by recognizing patterns of an intrusion. This paper outlines an investigation on the support vector machine models and choice of one of them for evaluation and implementation. The work also includes works on computer networks, providing a description and analysis of the structure of the computer network in order to generate network features.