| 一种新的RBF神经元网络分类算法 |
| |
| 引用本文： | 黄国宏,刘刚.一种新的RBF神经元网络分类算法[J].计算机仿真,2008,25(8). |
| |
| 作者姓名： | 黄国宏 刘刚 |
| |
| 作者单位： | 1. 广东工业大学信息工程学院,广东,广州,510006
2. 上海电力学院电力与自动化工程学院,上海,200090 |
| |
| 摘 要： | 为了改善对人工神经网络行为的认识和研究中的"黑匣子"式的难以处理的状态,基于RBF神经元模型的几何解释,提出了一种新的RBF神经网络分类算法,算法把RBF神经元看作是高维空间里的超球面,从而将神经网络训练问题转化为点集"包含"问题.同传统的RBF网络相比,算法能够自动地优化RBF网络中核函数的个数、中心和宽度,同时,省去了传统RBF神经网络输出层线性连接权的计算,简化了网络的学习过程,大大缩短了训练时间,并且通过实验证明了算法的有效性.
|
| 关 键 词： | 模式识别 神经网络 神经元 分类器 |
A New RBF Neuron Network Classification Algorithm |
| |
| HUANG Guo-hong,LIU Gang.A New RBF Neuron Network Classification Algorithm[J].Computer Simulation,2008,25(8). |
| |
| Authors: | HUANG Guo-hong LIU Gang |
| |
| Affiliation: | HUANG Guo-hong1,LIU Gang2(1.Dept.of Information,Guangdong University of Technology,Guangzhou Guangdong 510006,China,2.Dept.of Electric Power , Automation Engineering,Shanghai University of Electric Power,Shanghai 200090,China) |
| |
| Abstract: | In order to reduce the difficulty in the process of knowing and researching neural networks,based on the geometrical representation of RBF neuron model,this paper proposes a new classification algorithm,which regards the neurons of RBF neural networks as a series of hyperspheres in the high-dimension space.Thus the training problem of neural networks can be transformed into theincludingproblem of a point set.Compared with the classic RBF neural networks,the new algorithm has the ability of self-determining ... |
| |
| Keywords: | Pattern recognition Neural networks Neuron Classifier |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
