Improved Decision Tree Based On The Network Intrusion Detection

With the development of the network, security problems become more and more important. However, the original firewall has been unable to defend network alone, Intrusion Detection System plays an important role. Most current intrusion detection products are used by the simple pattern-matching technology, they exist in the consumption of resources, high rate of false positives, packet loss and other issues. Decision tree has many advantages: non-parametric, fast construction, and highly explanatory, these advantages have become widely used model in the data mining areas. This paper is on the main elements of the decision tree based on the intrusion detection system analysis, system design, the preliminary exploration.This paper first made the decision tree algorithm detailed analysis. According to the characteristics of the network data set's feature choice C4.5 algorithm. The C4.5 algorithm expanded the ID3 algorithm well, it will classify the domain to expand from the discrete type attribute values to the successive value attribute, it has become recognized in terms of the performance of the decision tree classification algorithm. In the analysis and summarizes the basic nature of C4.5 algorithm, performance and characteristics on the basis of this paper, the classical algorithm C4.5 some improvements, according to the characteristics of improved property values for attributes of the information gain, thereby enhancing the Decision Tree generation efficiency and improved analysis of the characteristics and effects.