A Method for Reducing the Number of Support Vectors in Fuzzy Support Vector Machine
Nguyễn Mạnh Cường, Nguyen Van Thien · Advances in intelligent systems and computing · 2016
We offer an efficient method to reduce the number of support vectors for Fuzzy Support Vector Machine. Firstly, we consider the Fuzzy Support Vector Machine model which was proposed by Lin and Wang. For the reducing the number of support vectors, we apply the $$l_0$$ regularization term to the dual form of this model. The resulting optimization problem is non-smooth and non-convex. The $$l_0$$ is then replaced by an approximation function. An algorithm which is based on DC programming and DCA is then investigated to solve this problem. Numerical results on real-world datasets show the efficiency and the superiority of our method versus the standard algorithm on both support vector reduction and classification.