SVM核函数的研究及其在语音激活检测中的应用

Research on Kernel Function of Support Vector Machine and Application to Voice Active Detection

在支持向量机的研究中,核函数起着关键性的作用。由于普通核函数都各有利弊,为了得到更好的学习和泛化能力,分别采用了混合核函数(mixture kernel)和适度衰减核函数(moderate decreasing kernel)。大量实验数据表明,将其应用于语音激活检测中,与其他几类分类核函数相比具有明显好的分类效果。

On the research of Support Vector Machine（ SVM）, It is important to choose an optimal kernel in order to enhance the characteristics of SVM. Since traditional kernel has its advantages and disadvantages,in our paper, we choose mixture kernel and moderate decreasing kernel which have better learning and generalization performance. Compared with other classifying kernels, a great deal of experimental results indicates that the new kernel has a better classifying performance in VAD （Voice Active Detection）.