Detection of novel attacks by anomaly intrusion detection system using classifiers

Nowadays analyzing unsuspicious network traffic has become a necessity to protect organizations from intruders. Really it is a big challenge to accurately identify threats due to the high volume of network traffic. In the existing system, to detect whether network traffic is normal or abnormal we need lots of information about the network. When lot of information is involved in the identification process the relationship between different attributes and the important attributes consider for classification plays an important role in the accuracy. Information gain selection process is used to provide a rank for features. Based on the rank, the most contributed features in the network is found and used to improve the detection rate based on the features selection. In this project, the performance of Lazy and Bayesian classifiers is analysed. In lazy classifier comes there are some algorithms namely, IBK and Kstar. Bayesian classifier comes there are some algorithms namely, Bayes Net, and Naïve Bayes. The performances of Bayesian and lazy classifiers are analysed by applying various performance metrics to identify the best classifier. It is observed that, the efficiency of lazy classifier is better as compared to that of Bayesian classifier.