The Classification Algorithm Based On I-K-Means Clustering And Na(?)ve Bayesian HRNB Classification Algorithm Is Used In Intrusion Detection

The development of high-technology led to the development of social information. Due to the rapid development of network technology,more and more people are starting to use the internet to deal with all kinds of things. At the same time,the Network Security questions also appropriately become the important guarantee of social development. Today is an era of information technology network. What firewall intercepts the internal ground attack is already exhausted. Not to mention to block some illegal attacks from outside. Network Security has become an important guarantee for the development of many technologies. Intrusion detection occupies a important position in Network Security. With the complex and diverse forms of network intrusion,Intrusion Detection System also must achieve a higher level to match the variation.Firstly,the thesis analyzes and introduce the intrusion detection. The basic principles of Bayesian are briefly introduced to the point. The shortcomings and drawbacks of intrusion detection using Bayesian algorithms are dissected. The main achievements consist of as following aspects:1).In order to overcome the traditional Na?ve Bayesian disadvantage for missing data,The HRNB Na?ve Bayesian classification algorithm is used to a hierarchical classification. So that the data set are divided into two groups:the complete set of attributes and missing attribute set categories. When the hierarchical process is being done, The “ ? ” named regulatory parameters is used to achieve the optimal classification.2).The original K-Means clustering algorithm is improved,in order to make the selection of initial values to avoid sensitive. This paper is using the Euclidean distance formula to calculate the degree of approximation of distance between class and class and intra class. The goal is to achieve maximum degree of approximation in class and minimum degree of approximation between class and class.3).The paper fusion the I-K-Means clustering algorithm and HRNB Na?ve Bayesian classification algorithm. Their respective strengths are combined。The new classification algorithm is proposed. The new algorithm is based on I-K-Means andHRNB Na?ve Bayesian. Intrusion Detection Model based on the new algorithm is established.The new algorithm are used in simulated experiment. Test with the data KDD Cup 10% is conducted The experimental results show that,In the intrusion detection for the data missing,the improved and new algorithm can increase the detection rate and reduce the omission rate and error rate. In intrusion detection of various attack types,the detection rate、omission rate and error rate are improved accordingly. It is proved that the new algorithm has certain validity and usability.