论文部分内容阅读
应用光滑函数改进支持向量机模型,得到无约束条件、可微的二次规划问题,从而可以采用快速的最优化算法求解光滑支持向量机模型.提出了一种广义三弯矩方法,用这个方法构造出新的五次样条光滑函数和七次样条光滑函数.证明了上述两个样条光滑函数的逼近精度均高于已有的各种光滑函数;基于上述两个样条函数的光滑支持向量机模型的收敛精度也高于已有的各种光滑支持向量机模型.
Applying the smoothing function to improve the support vector machine (SVM) model, the unconstrained and differentiable quadratic programming problem can be solved, so that the fast optimization algorithm can be used to solve the SLM model. A generalized three-moment method is proposed. A new five-spline smoothing function and a seven-spline smoothing function are constructed.It is proved that the approximation accuracy of the two spline smoothing functions is higher than that of the existing smoothing functions. Based on the smoothness of the two spline functions The SVM model also has higher convergence accuracy than the existing smooth support vector machine models.