| 归一化径向基函数网络的结构优化策略 |
| |
| 引用本文： | 祖家奎,曾庆丰,戴冠中. 归一化径向基函数网络的结构优化策略[J]. 计算机仿真, 2002, 19(3): 43-45,56 |
| |
| 作者姓名： | 祖家奎 曾庆丰 戴冠中 |
| |
| 作者单位： | 1. 西北工业大学,自动控制系
2. 西北工业大学,凝固技术国家重点实验室,陕西西安,710072 |
| |
| 摘 要： | 针对归一化RBF网络的结构设计，首先利用基于山峰函数的减法聚类算法构造网络的初始结构和初始数值，并采用奇异值分解（SVD）算法分析了网络初始结构中隐含层节点与奇异值，累积贡献率以及索引向量的关系，提出了针对归一化RBF网络的结构精简优化策略。最后，对该结构优化策略的可行性和有效性进行了仿真验证和性能比较。
|
| 关 键 词： | 归一化径向基函数网络 结构优化策略 减法聚类 奇异值分解 神经网络 |
| 文章编号： | 1006-9348(2002)03-0043-03 |
Structure Optimization Strategy of Normalized Radial Basis Function Networks |
| |
| Zu Jiakui ,Zeng Qingfeng ,Dai Guanzhong. Structure Optimization Strategy of Normalized Radial Basis Function Networks[J]. Computer Simulation, 2002, 19(3): 43-45,56 |
| |
| Authors: | Zu Jiakui Zeng Qingfeng Dai Guanzhong |
| |
| Affiliation: | Zu Jiakui 1,Zeng Qingfeng 2,Dai Guanzhong 1 |
| |
| Abstract: | Aimed at normalized radial basis function (NRBF) networks, this paper uses the subtractive clustering based on mountain functions to set up the initial structure of NRBF networks;adopts singular value decomposition (SVD) to analyze the relationship between neural nodes of the hidden layer and singular values, cumulative contribution ratio, index vector;then optimizes the structure of NRBF networks. Lastly, simulation and performance comparison is shown that the proposed algorithm is feasible and effective. |
| |
| Keywords: | Normalized radial basis function networks Subtractive clustering Singular value decomposition Structure optimization |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
