摘要
对支持向量机理论进行了简要分析,并将支持向量机引入汉语语音关键词识别系统中,根据关键词置信度将关键词假想命中分为接受和拒识两类,从而提高系统正确识别率。针对线性支持向量机、不同核函数下的非线性支持向量机以及核函数为径向基函数时支持向量机的性能做了一些相关实验。实验结果显示,支持向量机是一种相当有效的关键词确认方法。
Analyzed the theory of Support Vector Machine ( SVM ), the application of SVM technique for the problem of chinese keyword spotting is presented in this paper. The acceptance/rejection decision of a keyword is based on the confidence feature vector which is processed by a SVM classifier. Experiments are conducted by varying the kernel function and the results show that SVM is a useful method for keyword spotting.
出处
《计算机应用与软件》
CSCD
北大核心
2006年第7期85-87,共3页
Computer Applications and Software
关键词
支持向量机
关键词确认
核函数
Support vector machine Keyword verification Kernel function