Abstract: By considering the role of different samples on support vector machine, fuzzy support vector machine is presented based on support vector machine. However, it ignores the structural information of the given sample set. To this end, structural information of the given sample set is introduced into fuzzy support vector machine and obtained a structural fuzzy support vector machine model. It is converted to dual problem with quadratic programming using Lagrange method. Through solving this dual problem,the fuzzy support vector machine classifier is obtained. Experimental results in selected standard data sets demonstrate the effectiveness of the proposed method.

Key words: support vector machines, fuzzy support vector machine, structure information