| 基于RBFLN网络的改进RBF神经网络学习算法 |
| |
| 引用本文： | 韩敏,穆云峰.基于RBFLN网络的改进RBF神经网络学习算法[J].系统工程学报,2008,23(6). |
| |
| 作者姓名： | 韩敏 穆云峰 |
| |
| 作者单位： | 大连理工大学电子与信息工程学院,辽宁,大连,116023 |
| |
| 基金项目： | 国家自然科学基金，国家科技支撑计划，国家重点基础研究发展计划(973计划) |
| |
| 摘 要： | 提出了一种基于径向基链网络（RBFLN）的改进径向基函数（RBF）网络学习算法．网络结构采用RB—FLN模型，添加输入层对输出层的线性映射，在训练过程中基于最大误差学习样本对资源分配网络（RAN）新性条件进行改动，在不满足新性条件时，采用相似度参数对隐层中心和宽度进行调整；而满足新性条件时，对新增隐层节点也通过类均值的方法做出相应的改进．最后通过对无机建筑材料成分分析的仿真表明该算法可有效地简化网络结构，实现样本正确分类，并获得较好的校验能力．
|
| 关 键 词： | 径向基链网络 资源分配网络 最大误差样本 相似度 材料成分 |
Improved radial basis function neural network learning algorithm based on radial basis functional link nets |
| |
| HAN Min,MU Yun-feng.Improved radial basis function neural network learning algorithm based on radial basis functional link nets[J].Journal of Systems Engineering,2008,23(6). |
| |
| Authors: | HAN Min MU Yun-feng |
| |
| Institution: | School of Electronic and Information Engineering;Dalian University of Technology;Dalian 116023;China |
| |
| Abstract: | An improved RBF (radial basid function) neural network learning algorithm based on RBFLN (radial basis functional link nets) is proposed in this paper.The architecture of RBFLNI which includes a nonlinear model and an additive linear model due to extra linear mapping from the input to the output nodes,is used.The pattern which generates the maximal error is chosen to adjust the RAN (resource allocating network ) novelty during learning process.According to the novelty, the existing parameters of the network... |
| |
| Keywords: | |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
