Efficient intrusion detection system using random tree

Due to the rapid growth of the network technologies in today's society, new forms of network attacks are emerging and have made the computer network security a worldwide priority. Intrusion detection systems (IDS) methods are used for modelling and identifying normal and abusive system behaviour. IDSs employ misuse or signature-based and statistical anomaly models to detect intrusions. Due to large volumes of audit data as well as complex and dynamic properties of intrusion behaviour, optimising performance of IDS becomes an important problem. In this paper a novel method use the data mining technique like random tree for identifying intrusions is proposed. Random trees are an ensemble learning method for classification and regression that construct a number of decision trees at training time and produce the output that the mode of the classes output by individual trees. The experiment is evaluated on the KDD '99 datasets. The results show that the random tree algorithm achieves detection rates and false positive rates better than the existing system.