The Research On Intrusion Detection Based On Support Vector Machine

With the popularity of computer networks and applications, bringing greatconvenience to people’s lives at the same time, a growing number of network securityissues have emerged, people have become more emphasis on network-completeproblem. Because the traditional network security technologies, there are some flaws,widespread concern at home and abroad to protect network security intrusiondetection technology. How to quickly and efficiently identify attacks is the currentintrusion detection system pressing problems facing.Traditional intrusion detection methods prevalent false positives, false negatives,inefficiency and other shortcomings, can not meet the current needs of networksecurity. Current intrusion detection with high dimensional data is often smallsamples and inseparability. The support vector machine learning in small sampleclassification based learning methods developed to avoid the local optimal solution,and to overcome the dimension of the disaster. In solving small sample,high-dimensional input space intrusion detection classification problem has shownmany advantages thus support vector machine used in the field of intrusion detectionhas important theoretical and practical significance.This paper, statistical learning theory and support vector machine, whenanalyzed SVM learning algorithm is better than the other and discuss the reasons forSVM intrusion detection as well as the feasibility of support vector machine toachieve the basic idea of intrusion detection in SVM and discussed in detail theprinciples of intrusion detection based on the proposed model based on SVMintrusion detection and how to get the network packets. During the experiment, someof the experimental data and the characteristic values are discrete character, the firstcharacter of the experimental data are encoded, and then normalization, the originalexperimental data to support the identification number of vector functions vector, thenconducted experiments and analysis, experimental results show that the intrusiondetection using support vector machine can achieve a good detection rate.