Research On Algorithm And Its Application Based On Support Vector Machine

Support vector machine (SVM) is data mining technology based on statistical learning theory andstructural risk minimization, with the advantages of nonlinear, strong generalization ability, the globaloptimal and so on. However, at present there are still many problems unresolved. In this paper, thealgorithm of support vector machine is improved and applies it to practical problems. In this paper, themain work is as follows:First, the classification performance of support vector machine has a great relationship with theselection of kernel function. At present, the kernel function selection algorithm is numerous, but thereare still insufficient. In this paper, based on the analysis of shortcomings of previous methods, itproposes an algorithm of selecting kernel function based on improving the separability, and through theexperiment show that the method is effective.Second, support vector machine is unsatisfactory in the classification performance of minorityclass when dealing with imbalanced dataset. The traditional method of processing unbalanced datasetsoften ignore the influence of kernel function for classification performance. For this problem, analgorithm combining kernel function selection and under-sampling is presented in this paper.Experiments show that the method can effectively improve the classification performance.Third, the application of the support vector regression is studied in this paper, proposes a seawaterparameters prediction model based on SVR. The experimental results show that the proposed model isthe best in prediction performance compared with the traditional models, and it can more accuratelypredict the change tendency of seawater parameters. Being of the stronger theoretical foundation andhigher prediction ability, the proposed model can be applied to the practical seawater parametersprediction.