基于混合核函数的SVM在文本自动分类的应用

核函数是SVM的关键技术,核函数的选择将影响着学习机器的学习能力和泛化能力。不同的核函数确定了不同的非线性变换和特征空间,选取不同核函数训练SVM就会得到不同的分类效果。本文提出了一种混合的核函数1]Kmix=λKpoly+(1-λ)Krbf,从而兼并二项式核函数及径向基核函数的优势。实验证明选用混合核函数的支持向量机,与普通核函数构造的支持向量机的评估效果进行比较,混合核函数支持向量机具有较高的分类精度。

The Application of SVM in Automatic Text Classification Based on the Mixed Kernel Function

The kernel function is the key technology of the SVM,the choice of kernel function will affect the learning ability and generalization ability of learning machine.Different kernel functions determine the different non-linear transform and feature space,Select a different kernel function to train the SVM will get different classification results.In this paper,a hybrid kernel function Kmix=λKpoly+(1-λ)Krbf,the advantages of merger binomial kernel function and RBF kernel function.The experiments show that the selection of mixed kernel function support vector machine,compared to the combination of kernel function support vector machine has high classification accuracy assessment of the effect of the general kernel functions of support vector machines.