关键词: 稀疏性, 最小二乘支持向量机, 中心距离比, 边界样本

Abstract: To solve the problem of sparseness lacking in the Least Squares Support Vector Machine（LS-SVM），a new least squares support vector machine based on the boundary samples is proposed，which uses center distance ratio to select bigger support value boundary samples and making them as training samples.Thus，the number of support vector is reduced and the speed of computing is improved.Finally，the new algorithm is tested on the four benchmarking UCI datasets.The result shows that the proposed algorithm can adaptively obtain the sparse solutions almost not losing generalization performance，and the speed of classifiers is also improved.

Key words: sparseness, least squares support vector machine, center distance ratio, boundary sample