An improved cluster analysis algorithm using for network traffic flow

With the rapid development of computer network and the network application, network has plays an increasingly important role in the social progress and economic development. Rapid development of information technology makes the network traffic behavior has become increasingly complex, and the reliability of the network becomes crucial. Cluster algorithm using for network traffic flow is an entry to analysis network status. Support Vector Machine (SVM) is a machine learning method to solve binary classification problem. An improved cluster analysis algorithm of combining SVM with supervised subset density clustering is proposed in this paper, and minimize the training set of SVM by means of clustering is researched. A supervised self-adaptive method for the improved density clustering is designed to make out multiple centers choosing and referring the samples to SVM. The experimental results show that the algorithm reduces the iteration time of the whole training process without compromising the accuracy and generalization capacity of the algorithm obviously.