The fuzzy c-means (FCM) clustering algorithm is more suitable for intrusion detection, but the standard FCM does not consider the characteristics of each feature and the contribution rate to clustering analysis when calculating the distance between two samples, this obviously affects the authenticity and accuracy of the classification. Aim at the actual situation of intrusion detection data, a new weight calculation method is introduced in the paper, the method considers that there are independence factors exist for each feature of the sample, also the weight assignment of each feature should be related to the degree of its independence; while the independence degree of each feature depends on the cohesion and coupling of its value space. Simulation experiments shows that the new weight calculation method has higher classification accuracy, in practice it is very effective.
[1]
J. C. Dunn,et al.
A Fuzzy Relative of the ISODATA Process and Its Use in Detecting Compact Well-Separated Clusters
,
1973
.
[2]
Yadong Wang,et al.
Improving fuzzy c-means clustering based on feature-weight learning
,
2004,
Pattern Recognit. Lett..
[3]
LI Bai-nian.
Weigh on Cluster Fuzzy C-Mean
,
2007
.
[4]
Fan Jiu.
The New Explanation of Membership Degree in FCM and its Applications
,
2004
.
[5]
Meng Hai-dong.
Research based on euclid distance with weights of clustering method
,
2007
.
[6]
Wang Xin.
Construction of Fuzzy Similar Matrix
,
2003
.
[7]
James C. Bezdek,et al.
Fuzzy mathematics in pattern classification
,
1973
.