一类三阶多项式光滑的支持向量机  

A class of 3rd-order polynomial smooth support vector machines

作　　者：冯能山[1] 熊金志[1] 

机构地区：[1]东莞理工学院计算机学院,广东东莞523808

出　　处：《计算机工程与科学》2013年第6期156-161,共6页Computer Engineering & Science

基　　金：广东省自然科学基金资助项目(9151170003000017);东莞市科技计划资助项目(2012108102027)

摘　　要：研究了用一类新的光滑函数对支持向量机作光滑处理的问题,提出了一种新的具有三阶多项式光滑的支持向量机模型3SSVM,证明了该模型具有三阶光滑性和全局收敛性,并用Newton-Armijo算法进行了数值实验。结果表明,3SSVM模型的分类性能和计算速度优于PSSVM模型和SSVM模型,为光滑支持向量机的应用提供了更好的理论支持。Using a class of new smoothing functions, the problem of smoothing support vector ma- chines was studied. A new 3rd-order polynomial smooth support vector machine （3SSVM） model was proposed, and its 3rd-order smoothness and global convergence was proved. Numerical experiments were carried out to evaluate the 3SSVM by using Newton-Armijo algorithm. The results show that the 3SSVM model is better than the PSSVM model and the SSVM model, providing better theoretical sup- port for applications of smooth support vector machines.

关 键 词：分类 支持向量机 多项式 光滑函数 收敛 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]