中文摘要：

光滑函数用于支持向量机[1]可使原来不可微的模型变成可微的模型，从而可以采用快速的求解算法降低支持向量机的计算复杂性．2001年，Lee和Mangasarian用Sigmoid函数的积分函数作为光滑函数，提出了光滑的支持向量机模型（Smooth Support Vector Machine，SSVM）[2],SSVM具有很好的性质，是目前支持向量机领域的研究热点之一．

英文摘要：

Smoothing functions play an important role in Smooth Support Vec- tor Machine （SSVM） . The existing smooth approximation functions are almost all constructed based on polynomials, but known from the theory of rational approx- imation that polynomial function is not a good approximating tool for nonlinear function. In this study, the plus function x＋ = max{x, O} is first expanded into a power series, and based on this series and the theory of the formal orthogonal polynomials, an algorithm is given to compute a kind of Padd approximants. Then, based on Padd approximant, a new smooth support vector regression （Pade-SSVM） is achieved. Finally, theoretical analyses show that the Pad~-SSVM proposed in this study possesses better classification canahilltv than other existing SSVMs.