| 多输出神经元模型的多层前向神经网络及其应用 |
| |
| 引用本文： | 汪秉文,沈艳军,何统洲.多输出神经元模型的多层前向神经网络及其应用[J].控制理论与应用,2004,21(4):611-613. |
| |
| 作者姓名： | 汪秉文 沈艳军 何统洲 |
| |
| 作者单位： | 1. 华中科技大学,控制科学与工程系,湖北,武汉,430074
2. 陨阳师范高等专科学校,计算机系,湖北,陨阳,442700 |
| |
| 摘 要： | 研究了一种新的多输出神经元模型.首先,给出这类模型的一般形式,并将该模型应用于多层前向神经网络;其次,给出了其学习算法,即递推最小二乘算法,最后通过几个模拟实验表明,采用多输出神经元模型的多层前向神经网络,具有结构简单,泛化能力强,收敛速度快,收敛精度高等特点,其性能远远优于激活函数可调模型的多层前向神经网络.
|
| 关 键 词： | 神经网络 神经元模型 递推最小二乘算法 多输出神经元模型 |
| 文章编号： | 1000-8152(2004)04-0611-03 |
| 收稿时间： | 2003/2/19 0:00:00 |
| 修稿时间： | 2003/6/26 0:00:00 |
Multilayer feedforward neural networkswith multioutput neural model and its application |
| |
| WANG Bing-wen,SHEN Yan-jun,HE Tong-zhou.Multilayer feedforward neural networkswith multioutput neural model and its application[J].Control Theory & Applications,2004,21(4):611-613. |
| |
| Authors: | WANG Bing-wen SHEN Yan-jun HE Tong-zhou |
| |
| Affiliation: | Department of control science and engineering,Huazhong University of Science & Technology,Wuhan Hubei 430074,China; Computer Department,Yunyang Teachers'College,Yunyang Hubei 442700,China |
| |
| Abstract: | A multi_output neural model and its general form are presented.The recurrent least square,a learning algorithm,was used to train multilayer feedforward neural networks(MFNN) with this new model.Several simulations demonstrated that MO (multi_output neural)_MFNN has simple architecture, excellent generalization capacity, fast speed of convergence and improved accuracy. Its performance is superior to TAF (tunable activation function) MFNN. |
| |
| Keywords: | neural networks neural model recurrent least square(RLS) multi_output neural model |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
| 点击此处可从《控制理论与应用》浏览原始摘要信息 |
|
点击此处可从《控制理论与应用》下载全文 |
|
