Spline Polynomial Smooth Support Vector Regression

摘要

Smooth function is the essence of smooth support vector machine, which has been studied in classification widely. Smoothing technology can be also applied in support vector regression. This paper proposes spline polynomial function as a new kind of smooth function for support vector regression. We introduce an instance of this class of functions and give the theoretical proof of its advantage on efficiency and precision over other smooth functions. For support vector regression, we develop a new model of Smooth Support Vector Regression with the new smooth function.