Network situation prediction based on optimized SVR model
Yu Feng, Wei Liu, Gao Chunyang, Bai Liang · 2013
Accurate assessment of network situation has important function in improving the performance and QoS of the network. Since support vector regression emerges as a novel machine study method, it is good at searching implied regression functions according to limit observing data. SVR functions can be used to predict future data. Although PSO-based algorithms are easy, they tend to drop into local extremum. Therefore, chaos particle swarm optimization is used in this paper to optimize the vector parameter, based on which the network situation predicting model is established. The experiments adopt the dataset in Honeynet and actual network traffic to verify the effect of improved algorithm. The results show that chaos PSO scheme has overcome the subjectivity in SVM parameters selection effectively. It praised the prediction accuracy of network situation and has better comprehensive performance compared to traditional prediction methods.