摘要
在目标识别问题中，一类样本可能具有两个或更多的聚类中心，运用Ｂａｙｅｓ算法会产生较大误判率。本文采用Ｋｏｈｏｎｅｎ算法有效地解决了这一问题并对此进行了仿真和详细的数学分析，说明了Ｋｏｈｏｎｅｎ网络在解决此类问题中的优越性。
In target recognition problems, one cluster of patterns has two or more cluster centers. This problem sloved with Bayes algorithm gets some unnecessary decision error. In this paper it promotes Kohonen self organization model and gets good simulation results. Then a thorough analysis for this problem is made and the advantage for Kohonen algorithm is proved.
出处
《数据采集与处理》
CSCD
1999年第2期258-261,共4页
Journal of Data Acquisition and Processing
关键词
目标识别
神经网络
模式识别
KOHONEN网络
target recognition
learning algorithm
neural network
Kohonen self organization model
