| 一种用于模式识别的动态RBF神经网络算法 |
| |
| 引用本文： | 韩敏,崔丕锁. 一种用于模式识别的动态RBF神经网络算法[J]. 大连理工大学学报, 2006, 46(5): 746-751 |
| |
| 作者姓名： | 韩敏 崔丕锁 |
| |
| 作者单位： | 大连理工大学,电子与信息工程学院,辽宁,大连,116024;大连理工大学,电子与信息工程学院,辽宁,大连,116024 |
| |
| 摘 要： | 对径向基函数（RBF）神经网络在数据分类中的应用进行了研究．提出一种应用于模式识别的动态RBF训练算法，该算法使用区域映射误差函数并结合资源分配网络（RAN）的“新性”（novelty）条件动态调整网络的隐层节点数，从而可以更加有效地进行模式识别．二分类样本和建筑材料CaO-Al2O3-SiO2系统仿真表明，该改进算法使误差下降更快，减少了训练次数，可以获得精简的网络结构，从而使网络具有较高的泛化能力．
|
| 关 键 词： | 径向基函数 分类 区域映射 模式识别 |
| 文章编号： | 1000-8608（2006）05-0746-06 |
| 收稿时间： | 2005-03-19 |
| 修稿时间： | 2005-03-192006-08-20 |
A dynamic RBF neural network algorithm used in pattern recognition |
| |
| HAN Min,CUI Pi-suo. A dynamic RBF neural network algorithm used in pattern recognition[J]. Journal of Dalian University of Technology, 2006, 46(5): 746-751 |
| |
| Authors: | HAN Min CUI Pi-suo |
| |
| Affiliation: | School of Electr. and Inf. Eng., Dalian Univ. of Technol., Dalian 116024, China |
| |
| Abstract: | The application of radial basic function(RBF) neural network in the data classification is studied.A new dynamic training algorithm for RBF network used in pattern recognition is proposed.It uses the regional mapping error function and the novelty condition of the resource-allocating network(RAN) to dynamically adjust the nodes in the hidden layer of the network,and makes the pattern recognition more efficient.By the simulation result of the modeling of synthetic two-class problem and CaO-Al_2O_3-SiO_2 system,it is proven that the algorithm can make the descending speed of error more quick and shorten the training times,thus the network with the concise structure is obtained and better generalization is achieved. |
| |
| Keywords: | RBF classification regional mapping pattern recognition |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
