元音识别的几种分类模型的比较与分析  

Comparison and Analysis of Vowel Recognition Classification Models

作　　者：陈志辉[1] 吴敏敏[1] 

机构地区：[1]莆田学院电子信息工程系,福建莆田351100

出　　处：《莆田学院学报》2012年第5期62-65,共4页Journal of putian University

基　　金：福建省科技厅重大基金项目(2010H6019)

摘　　要：根据元音数据处理方法和选取的特征,提出了三种基于统计学的模式分类方法:线性判别、二次判别和k-最近邻,并通过模型选择方法,对实际数据进行测试与比较,得出k-最近邻分类模式最为合适,同时提出进一步改进的思路。According to the features of vowels data processing and selection, the article presented three classification methods of patterns based on statistics theory, such as linear discriminant, quadratic discriminant and k-nearest neighbor. And through the model selection concluded that k-nearest neighbor was the most suitable still needed in this model. method, actual data of tested and compared, it was classification pattern but further improvement was

关 键 词：元音数据 分类方法 线性判别 二次判别 K-最近邻 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] TN912.34[自动化与计算机技术—计算机科学与技术]