| Artificial Immune is an increasingly important area in the field of intellective technology. Biological Immune is a highly complexity and self-adaptive system with capability of learning,memory acquisition,pattern recognition and so on. By simulating and applying the biological immune system,the new computational techniques can solve not only the science but also the engineering problems.This paper does research on a sub-item of the project sponsored by National Science Fund.Simulating some conception and theory about immune in life sciences,and focusing our mind on the technologies such as immune evolution,antibody clone selection and so on, the tasks of this item are doing deep research on applications of data-mining and apply it to a network intrusion detection system.The main work and achievements are as follows.(1)A new association rule mining algorithm is proposed,which is based on an advanced artificial immune alrorithm(AIA).This algorithm mines the association rules of transaction database directly, according to the attributes which users are interested in.Then,the advantages and disadvantages of the new algorithm are analysed and discussed.(2)To meet more extensive mining demand,this paper presents another immune association rule mining method,which aimes at the largest frequent item set.The primary idea of this method is introducing the Immune Clone Selection Theory into the mining process.This method can reduces the cost of scanning the database.At the same time, its confidence converges rapidly.(3)By simulation experiment,we compare the last two algorithms with the traditional Apriori algorithm and association rule mining algorithm based on evolution,study their differences in capability,and analyse their advantages and limitations.(4)Finally,we apply the data mining algorithm based on Clone Selection Theory to an appropriate field——data mining in network intrusion detection system.We can learn that lower support makes this algorithm more efficient.The result of... |
PDF Full Text Request