Propose NB/HNB Classifiers to Build NIDS


This paper indicates that the potential attack to traditional/cloud network is Denial of Service (DoS) attack that effect on the availability of the resource, to solve this problem; this paper propose hidden naïve bays(HNB) classifier to enhance the accuracy of detect DoS attack in cloud network with taking into consideration the traditional environment, the system applied NB classifier firstly supported by discretization and feature selection method to show the difference between the traditional NB classifier and the new model HNB classifier. Two methods are used to select the best feature (Info Gain and Gain ratio) and by used two dataset (KDD cup 99 and NSL KDD datasets) that are used to evaluate the performance of the system. The experiential result show that the proposed system based on HNB classifier enhance the accuracy of detect DoS attack where reach to 100% in three test dataset that are different in size and content by use KDD cup 99 dataset and select only twelve features depended on gain ratio as feature selection, while when used NB classifier the accuracy of detect DoS is equal (94, 97, 98) in three different test dataset. In NSL KDD dataset the accuracy of detect DoS reach to 90% for three test dataset based on HNB classifier and by select 10 features based on GR method, while when used NB classifier is equal to (88, 87, 86) for three test dataset.