Abstract: The BFGS-Armijo method and Newton-Armijo method are two popular methods to solve the smooth models. The authors respectively listed the procedures of these two methods to solve Smooth Support Vector Machine (SSVM), and made a comparative study between the two methods to solve SSVM. The numerical results show that the classification performance of BFGS-Armijo method is almost the same as the Newton-Armijo method, but the classification efficiency of Newton-Armijo method is 26.2% higher than that of BFGS-Armijo method.

摘要： BFGS-Armijo法和Newton-Armijo法是求解光滑模型的常用算法。分别列出用此两种算法求解光滑支持向量机模型(SSVM)的具体步骤,并用这两种算法对求解SSVM模型进行比较研究。数值实验结果表明：Newton-Armijo法的分类性能和BFGS-Armijo法基本相同,而分类效率比BFGS-Armijo法高出约26.2%。

关键词: 分类, 支持向量机, Newton-Armijo法, BFGS-Armijo法, classification, Support Vector Machines (SVM), Newton-Armijo method, BFGS-Armijo method