| 一种自适应的小波神经网络 |
| |
| 引用本文： | 黄凤岗,孙文彦,玉莹. 一种自适应的小波神经网络[J]. 电子学报, 1998, 0(8) |
| |
| 作者姓名： | 黄凤岗 孙文彦 玉莹 |
| |
| 作者单位： | 哈尔滨工程大学计算机与信息科学系!哈尔滨，150001，哈尔滨工程大学计算机与信息科学系!哈尔滨，150001，哈尔滨工程大学计算机与信息科学系!哈尔滨，150001 |
| |
| 摘 要： | 本文受感知域划分思想的启发，将小波的多分辨分析与BP网结构相结合，构造了一种新的小波神经网络．该小波神经网络利用多分辨分析生成小波树，小波树的生长与网络的训练相结合，自适应地生成隐层结点，并且删除分类不佳的结点．以声纳信号进行了实验，结果表明：该网络充分发挥了小波的特点，将模式识别的特征抽取与分类器设计融为一体．
|
| 关 键 词： | 小波神经网络 多分辨分析 BP网 小波树 |
An Adaptive Wavelet Neural Network |
| |
| Huang Fenggang, Sun Wenyan, Wang Ying. An Adaptive Wavelet Neural Network[J]. Acta Electronica Sinica, 1998, 0(8) |
| |
| Authors: | Huang Fenggang Sun Wenyan Wang Ying |
| |
| Abstract: | Enlighted by RFP(Receptive Field Partitioning), we construct a new WNN which conbines multiredution and BP neural network structure. The WNN uses multiredution to construct a wavelet tree. The training of network is combined with the growth of the wavelet tree, so that the hidden nodes can be constructed adaptively,and nodes which are not gial at Pattern classification can be deleted. We also do experiments on active sonar signal. Experimental results show that the WNN can integrate pattern abstraction and classifier design. |
| |
| Keywords: | Wavelet neural network Multiresolution BP network Wavelet tree |
| 本文献已被 CNKI 等数据库收录！ |
