Research on Incremental Learning of SVM Based on Robustness

Abstract

The following problems exist in the process of incremental learning of support vector machines. If the number of support vectors will increase with the increase of the increment sample, the time of training will become longer and longer; if the vector that has no effect on the hyperplane of the classification is abandoned, this part of the vector may become a support vector in the subsequent training. This will have an impact on the effect of the classification. In this paper, a support vector machine incremental learning algorithm based on fuzzy C mean clustering and central density is used to determine support vector set by fuzzy C mean clustering, and the selected non support vector sets are obtained by using the ratio of center density to determine non support vectors. The effect of non support vector on the robustness of support vector incremental learning is studied by comparing the classification efficiency of the four kinds of classification, and the conclusion of this paper is finally obtained.