Research of the Intrusion Detection Based on Weighted Features Unsupervised Fuzzy Clustering
Xinyu Li · Natural Science Journal of Xiangtan University · 2011
A distance measurement method based on weighted features of heterogeneous data which is designed to response for network intrusion detection traffic data samples heterogeneity of Characteristics properties and differences of contribution rates between properties has been put forward in this paper.For the Fuzzy C-means clustering algorithm for network intrusion detection problem is difficult to determine the number of clustering,this paper proposes the unsupervised fuzzy clustering algorithm for intrusion detection that automatically determine the optimal clustering number.These results of simulations run on the KDDcup1999 data sets show that the algorithm has efficient performance in clustering number optimization and effectively increases the detection degree of clustering intrusion detection.