Application Research On Data Mining To Intrusion Detection

With the rapid development of computer hardware and network technology, information has become the key factor in promoting the development of economy and society. When people share resource with others, they also found that problem of information security become increasingly prominent. Network attacks are becoming increasingly multiplex, sophisticated and intelligent; only depend on the traditional static security defense technologies can not meet the network security needs. As a typical active defense technology intrusion detection technology has become an important tool in protecting network security.The intrusion detection system construct on traditional hand-coding has some limitations such as detection accuracy; adaptability and scalability. Data mining can be used in intrusion detection system to solve these problems. At first the paper introduces the concepts of intrusion detection and data mining, common intrusion detection techniques and models, analyze the application of data mining techniques in the detection system. On this basis, the paper expand the current architecture of intrusion detection systems, presents a data mining-based intrusion detection system model, then design the model works, processes, and the details of functional modules. The model uses clustering algorithms to categorize the data set, and models do not need marked training data set, so this model has certain flexibility and expansibility.This paper focuses on the key algorithm of Apriori in data mining. According to the statistical nature of database transaction, this paper proposes an algorithm for discovering maximum frequent itemsets that based on layer. First of all, the algorithm judge the overall frequency of the candidate itemsets, and then through the introduction of the overall strategy, sequencing strategy, the minimum strategy effectively reduces the comparison times between database transactions and candidates in the process of discovering maximum frequent itemsets. Experiments show that the algorithm is more effective than Apriori algorithm.