摘要
基于人的大脑进行聚类分析所遵循的基本原则,提出了一种模糊超球质心聚类神经网络学习算法。该方法无需用户事先给定聚类个数K,通过神经网络自组织学习,可以正确识别聚类个数与聚类中心。实验结果表明,该算法是一种全新的聚类方法,具有学习时间短,稳定性强且不依赖于聚类样本的输入顺序等优点。
Based on the fundamental rules of human brain's analyzing and reasoning in clustering, this paper proposes a learning algorithm of neural networks of fuzzy hypersphere barycenter Clustering. Clustering numbers need not be provided for the algorithm. Through self-adapt learning of neural networks, the clustering numbers and centers can be established correctly. Simulation results indicate that the algorithm is a totally new clustering method, which is characterized by high efficiency, strong stability and independence from input order of clustering samples.
出处
《辽宁工程技术大学学报(自然科学版)》
CAS
北大核心
2001年第5期656-658,共3页
Journal of Liaoning Technical University (Natural Science)
关键词
模糊超球隶属函数
质心聚类法
神经网络
学习算法
fuzzy hypersphere membership function
barycenter clustering method
neural networks