Research on multi-classification algorithm for Semi-supervised Support Vector Data Description

This paper describes the classification and characteristics of single-classification support vector machine, and the advantage applied it to solve the multi-classification; then, combining the algorithm based on support vector data field description with semi-supervised learning idea, propose a semi-supervised support vector data field description multi-classification learning algorithm. This algorithm determine accept the label and refuse the label by defining the membership of non-target samples; through constructing more super ball on the target sample set and the labeled non-target sample set, realize the multi-classification algorithm based on support vector data field description.