Intrusion Detection Research Based On C4.5Decision Tree Algorithm

With the rapid development and wide application of computer and networktechnology, sharing information from the network resources has become one of theessential ways in people’s daily life. At the same time, people also have to face a series ofnetwork security problems caused by the invasion. The traditional firewall technology hasbeen difficult to protect network security separately. Intrusion detection begins to play anirreplaceable role as a complement of firewall technology.Intrusion detection is a security technology to discover intrusion attacks through areal-time monitoring of the target system. Most of the current intrusion detection systemsuse a simple rule-based pattern matching techniques. They have some disadvantages thatthey have a large calculation and high rate of false positives and omission. For thesedeficiencies, this paper proposes a decision tree method as a descriptive model，usingMATLAB to program the C4.5algorithm to build a classification tree with the training setto detect the intrusions. This algorithm which has some characteristics of fast construction,high classification accuracy, high detection rate and good adaptive and self-learningperformance is suitable for intrusion detection.This article briefly describes the background of intrusion detection and discusses itsresearch status, classification and development trends first. Then it gives a briefintroduction to data mining technology. Next, it focuses on introducing several majoralgorithms. This work selects the C4.5algorithm based on the characteristics of networkdataset. It creates a decision tree after analyzing and summarizing the basic natures,performances and features of the C4.5algorithm. Finally, the article gives the experimentalprocess of the application of C4.5algorithm and the simulation results of the IDS.