| 基于人工神经网络的聚类算法 |
| |
| 引用本文： | 陈华,;管乐乐,;陈秉岩,;刘志军.基于人工神经网络的聚类算法[J].淮南工业学院学报,2009(2):38-41. |
| |
| 作者姓名： | 陈华 ;管乐乐 ;陈秉岩 ;刘志军 |
| |
| 作者单位： | [1]河海大学常州校区数理部,江苏常州213022; [2]河海大学常州校区商学院,江苏常州213022; [3]河海大学常州校区计算机与信息学院,江苏常州213022 |
| |
| 基金项目： | 河海大学常州校区青年科技基金资助项目（XZX/08B001--05）;国家大学生创新训练计划资助项目（C2008025）;学生科技基金资助项目（08923）作者简介： |
| |
| 摘 要： | 研究连续型Hopfield神经网络的电路机理,推导出网络的权值计算公式,并运用连续型的神经网络模型构造出聚类算法;对20个随机生成数据计算模拟神经网络的运行,逐步调整类内精度多次训练,当参数a=b=500,c=200且类内精度控制不超过0.6时,有一个优化的聚类方案输出。
|
| 关 键 词： | Hopfield神经网络 聚类 连接权 |
Clustering Algorithm Based On Artificial Neural Network |
| |
| Affiliation: | CHEN Hua ,GUAN Le-le,CHEN Bing-yan ,LIU Zhi-jun (1. Department of Mathematics and Physics, Changzhou Campus of Hohai University, Changzhou Jiangsu 213022, China; 2. Department of Business Administration, Changzhou Campus of Hohai University, Changzhou Jiangsu 213022, China; 3. Department of Computer Seience,Changzhou Campus of Hohai University, Changzhou Jiangsu 213022, China) |
| |
| Abstract: | Electric circuit mechanism of continual Hopfield neural network was discussed. The formula of network right weights calculation was educed. The clustering algorithm was established by continual neural network model. 20 random test data were input to simulate operation of the neural network, which is educated many times by gradual regulation of precision in species, and optimal clustering solution output was obtained when parameters a =b= 500,c= 200,and precision control in species not more than 0. 6. |
| |
| Keywords: | Hopfield neural network clustering connection weight |
| 本文献已被 维普 等数据库收录！ |
|
