基于区间值模糊逻辑神经元的区间值模糊C-均值聚类神经网络  

Interval-Valued Fuzzy C-Means Clustering Neural Networks Based on Interval-Valued Fuzzy Logic Neurons

作　　者：李文化[1] 王智顺[2] 谢维信[3] 何振亚[2] 

机构地区：[1]烟台东方电子集团,烟台264001 [2]东南大学无线电工程系,南京210096 [3]西安电子科技大学,西安710071

出　　处：《电子学报》1998年第10期99-103,共5页Acta Electronica Sinica

基　　金：国家攀登计划认知科学(神经网络)重大关键项目;国家自然科学基金

摘　　要：本文提出了一种基于区间值模糊逻辑神经元的三层前馈自组织神经网络模型，用来实现区间值模糊C-均值聚类分析.网络第一、二层神经元的输入、输出和权连接取值属于区间值模糊集I[0，1].第一层神经元为区间值线性神经元;第二层为区间值模糊相等神经元，其功能是实现输入样本与各类的匹配运算.本文采用区间值模糊相等关系作为匹配的指标为了定义区间值模糊相等神经元，本文在点值模糊相等关系的基础上推导了区间值模糊相等关系的计算方法;第三层神经元为模糊竞争神经元，各神经元的输出代表输入样本的模糊分类结果此外.本文提出了一种区间值模糊竞争学习算法用于区间值模糊C-均值聚类神经网络的训练.In this paper a three-layered feedforward self-organizing neural network model is proposed based on interval-valued fuzzy logic neurons in order to realize the interval-t,alued fuzzy C-means (IVFCM) clustering analysis. The inputs/outputs of the first and second layer neurons and weights between them belong to foe interval-valued fuzzy sets I [0, 1]. The neurons in the first layer are intervalvalued linear neurons; The neurons in the second layer are interval-valued fuzzy equality neurons (IVFEN) which realize the matching computation between the input samples and all of the clusters. Interval-valued fuzzy equality relation is used to compute matching. In order to define the model of IVFEN, the computation method of interval-valued fuzzy equality relation is deduced based on point-valued fuzzy equality relation. The neurons in the third layer are fuzzy competition neurons. The outputs of the third-layer neurons represent the fuzzy classification result of the input sample. Furthermore, intervalvalued fuzzy competition learning (IVFCL) is proposed to train the IVFCM neural network (IVFCMNN)

关 键 词：C-均值聚类 自组织神经网络 区间值模糊 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]