Research On Technology Of Intrusion Detection Based On Improved Naive Bayesian Algorithm

With the coming of information age, the Internet has been continuously towards a broader scale, faster development. Computer network for the whole of the human society has brought the unprecedented change, it affects people’s work, entertainment,and every aspect of life. Today, globalization of the Internet has penetrated into various fields such as politics, economy and culture, and class for each industry to provide reliable and convenient service. However, Internet as "double-edged sword" to create benefits for human beings, the underlying network security hidden trouble at the same time. These hazards to a threat to privacy leaks, damage to property, political and social unrest and other major hazards. So, the problem of network security research and development has become the attention hot spot of globalisation.According to the current development situation of network security technology,this paper launches the research of network intrusion detection technology, through summarizing and analyzing the previous research results, aiming at the problems that exist in the intrusion detection technology, the introduction of data mining algorithm,Naive Bayesian technical train of thought, put forward the ITWNB intrusion detection model, this model is the extension and improvement of the current intrusion detection technology, its technical innovations and work content is as follows:Firstly, through the analysis of the shortcomings of traditional NB algorithm, this paper puts forward the “ITS” based on decision tree theory, through the idea of weighted to relax the limitation of the “conditional independence assumption”, the model to improve the classification performance of the NB algorithm.Secondly, “ability correction algorithm” is proposed in this paper to constantly revised to build “information tree” the misclassification of sample collection, by modifying the maximum a posterior probability approach to decision-making or a tendency to further improve the classification accuracy of NB algorithm.Thirdly, After the above improvement of NB algorithm based on improved intrusion detection model is established, based on the ITWNB algorithm in intrusion detection model, and combines the data of “a semi-supervised learning” training mode,improve the efficiency of intrusion detection.Finally, this paper used 10% of the subset of the KDD ’99 intrusion detection data set as a data source for intrusion detection model is proposed in the experiment test,through the experimental data of results to verify the feasibility of the model and compared with the other Bayes algorithms in intrusion detection model show that the superiority and stability of the proposed model.