模糊中心聚类的模式识别学习方法

Pattern Recognition With Fuzzy Central Clustering Algorithms

基于一个约束条件下的非线性规划问题的优化计算思想 ,把模糊中心聚类中计算输入矢量与中心的距离来实现聚类作为一种优化计算问题 ,证明了模糊中心聚类方法 ,取一个适当的属函数 ,其聚类中心vi 为模糊聚类中心价值函数的极小值 ,推导出了基于模糊中心聚类的模式识别的无导师递推学习方法 ,提出了模糊中心聚类模式分类神经网络结构 。

Based on optimization of constrained nonlinear programming, an approach of clustering center and a fuzzy membership function of pattern classification are derived from an objective function of the constrained nonlinear programming. An unsupervised algorithm with recursive expression and a fuzzy central cluster neural network are suggested in this paper. The fuzzy central cluster neural network proposed here can realize crisp decision or fuzzy decision in pattern classification.[