Optimized k-means clustering algorithm based on artificial fish swarm

To improve the situation of insufficient global research in K-Means Algorithm, this paper discusses an optimized K-Means clustering algorithm based on artificial fish swarm, which overcomes the sensitivity to initial clustering center selection of K-Means clustering algorithm and gets the optimal global clustering partition. Meanwhile, to improve the precision of clustering algorithm, a novel algorithm is presented to calculate the inner-class distance and inter-class distance. Simulation experiments have been implemented over data set KDD-99, and the results showed that the satisfactory detection rate and false acceptance rate could be obtained in network intrusion detection.