Fuzzy support vector machines based on collaborative representation

The design of fuzzy membership functions is the key to fuzzy support vector machines. The fuzzy membership functions of traditional fuzzy support vector machines are based on the distances between the samples and the class center. However, it is easily affected by the sample points distribution. To alleviate this problem, in this paper, we proposed a new design of fuzzy membership function. The new method calculated fuzzy membership values based on the collaborative representation coefficients among the sample points. Experimental results show that the new method achieves higher accuracy which verifies its effectiveness.