摘要
通常我们用K -平均法和K -邻近法估计椭圆基函数 (EBF)中心位置与函数宽度等参数。但上述的方法在输入矢量包含相关元素时存在性能次优化问题。另外 ,对于EBF网络来说 ,如何选择适当的类的数目仍是一个难以解决的问题。本文提出用结合改进的RPCL算法和EM算法的EBF网络结构来解决上述问题。在话者识别的软件开发中 。
The use of the K -means and the K -nearest neighbor heuristic in estimating the elliptical basis function(EBF)parameters may produce sub-optimal performance when the input vectors contain correlated components.And how to determine an appropriate number of clusters of EBFN is also a problem.This paper proposes to overcome those problems by incorporating the improved RPCL (rival penalized competitive learning)algorithm and the EM(expectation maximization)algorithm into the EBF structure.A speaker recognition software based on the idea above was developed.And it shows this method has a better representation of data set and a lower verification error rate.
出处
《计算机应用与软件》
CSCD
北大核心
2003年第9期3-5,共3页
Computer Applications and Software
基金
上海市科学技术发展基金项目 (编号沪科鉴 0 0 2G0 0 0 2 6)
