论文部分内容阅读
针对核函数缺少统一范式这一问题,以及为进一步提高支持向量机的性能,基于Askey-Wilson多项式提出一类新的支持向量机核函数,即AW核函数,从而为支持向量机核函数提供一种新的选择.首先,基于Askey-Wilson多项式参数的特性对其进行化简,得到除阶数外仅包含一个参数的AW多项式,以避免引入更多的核参数;然后,根据AW多项式的正交特性以及核函数的性质,构建基于AW多项式和RBF核函数的新核函数.实验结果表明,AW核函数能有效提高支持向量机的分类性能.
Aiming at the problem that the kernel function lacks a unified paradigm, and to further improve the performance of the support vector machine, a new kind of kernel function of support vector machine (AW kernel function) is proposed based on the Askey-Wilson polynomial to provide a support vector machine kernel function We first simplify AWkey polynomials based on the properties of Askey-Wilson polynomial to get an AW polynomial containing only one parameter in addition to the order to avoid introducing more nuclear parameters. Then, according to the positive polynomial of AW polynomial And the properties of kernel function, a new kernel function based on AW polynomial and RBF kernel function is constructed.The experimental results show that AW kernel function can effectively improve the classification performance of SVM.