On Classification Method Of Network Flow Based On Multi-core Support Vector Machine (SVM)

With the development of network technology and the network popularization, theamount of information on the network is more and more big, the classification ofnetwork traffic transfer, will reduce the pressure of the network transmission, increasethe speed of the network transmission. This article mainly research network trafficclassification method based on vector machine (SVM), the fundamental nature ofnetwork traffic classification is based on the packet length to classify.Previous way of classification of network traffic is assigned according to the staticport. Of this study is based on vector machine (SVM) classification of network, supportvector machine (SVM) classification is based on a statistical basis for classification is akind of dynamic classification method and support vector machine method has strongability of analysis and near capacity. Vector machine (SVM) is a research way of neuralnetwork, the application of this approach to classify network traffic has a very goodadvantage, so this way is more and more attention, this way has the ability of selflearning, has the very good intelligence and learning. This method can classify thenetwork data.Network traffic classification based on vector machine broke the traditional way ofstatic classification in port, the dynamic network traffic are classified, to ensure that thenetwork quality of service and network transmission speed, to improve the precision ofthe classification of network traffic.This paper analyzes the mixed together nuclear support vector machine (SVM)algorithm, etc. and attempts to ensure that the network quality of service, as well as thetransmission speed. This paper also analyzed the MKSVM the role of the Internet trafficclassifier, and the application of the interface and the result of data analysis, designedthe analysis system of vector machine test and the interface design of the whole system,the transplant and classification module are analyzed, and design the whole analysissystem, the network classification method based on vector machine carried on thethorough research.