双重模糊K-均值算法的分类器设计  

Design of Dual Fuzzy K-means Classifier

作　　者：李泰[1] 沈祥红[1] 

机构地区：[1]商丘职业技术学院,河南商丘476000

出　　处：《计算机测量与控制》2008年第9期1325-1326,1334,共3页Computer Measurement &Control

摘　　要：为了提高分类器的分类率,再一次把模糊的思想引入K-均值算法,构成双重模糊K-均值算法的分类器,所不同的是把模糊化思想引入到分类规则上;用这样一个模糊规则来表示分类的模糊系统,更加有效地构建了一个能够对训练样本比较准确分类的模糊分类器,用这种方法设计的分类器有效地提高了分类器的分类率;最后用Iris数据进行仿真测试,测试结果显示其分类率能够达到98%左右,并且不需要预定义参数,训练时间短,方法简单。For increasing classifier＇ s classification rate, We make use of the fuzzy theories to K-means algorithm again . This time the fuzzy theories were led into classification rule. The fuzzy system design method that uses a fuzzy rule to represent a cluster is then propsed so that a fuzzy classifier can be efficiently constructed to correctly classify the training data. And the design method increased effi ciently classifier＇ s classification rate. Finally, we test the classifier with the Iris data, Test display the classification rate is about 98%. The proposed method does not need a prior parameter definition, but only needs a short training time, therefore is simple.

关 键 词：模式识别 模糊分类器 K-均值算法 

分 类 号：TP391.44[自动化与计算机技术—计算机应用技术]