| 自适应模糊神经网络构建方法 |
| |
| 引用本文： | 马波,王正群,沈杰,邹军. 自适应模糊神经网络构建方法[J]. 江南大学学报(自然科学版), 2007, 6(6): 821-824 |
| |
| 作者姓名： | 马波 王正群 沈杰 邹军 |
| |
| 作者单位： | 扬州大学,信息工程学院,江苏,扬州,225009 |
| |
| 基金项目： | 江苏省高校自然科学基金项目(05KJB5201)，扬州大学自然科学基金项目(KK0413160) |
| |
| 摘 要： | 模糊极大极小神经网络的结构与超盒形状系数有关,该神经网络的性能取决于超盒形状系数的选择.在构建该神经网络时,最优超盒形状系数的确定比较困难,故提出了一种自适应的模糊极大极小神经网络构建方法,取消了超盒形状系数对扩张过程的限制,以是否包含其他类样本为超盒扩张条件.实验结果表明,使用这种模糊神经网络方法生成的神经网络结构更简单,对模式分类的效果更好.
|
| 关 键 词： | 神经网络 模糊神经网络 模式识别 |
| 文章编号： | 1671-7147(2007)06-0821-04 |
| 修稿时间： | 2007-06-15 |
An Adaptive Construction of Fuzzy Neural Network |
| |
| MA Bo,WANG Zheng-Qun,SHEN Jie,ZOU Jun. An Adaptive Construction of Fuzzy Neural Network[J]. Journal of Southern Yangtze University:Natural Science Edition, 2007, 6(6): 821-824 |
| |
| Authors: | MA Bo WANG Zheng-Qun SHEN Jie ZOU Jun |
| |
| Abstract: | Structure of fuzzy max-min neural network is related to hyper-box's shape factor,which also decides its performance.But It is difficult to get the optimum hyper-box's shape factor.A method of constructing adaptive fuzzy max-min neural network is given in this paper.The method cancels the restriction of hyper-box's shape factor for expanding process,and take containing other kinds of samples or not for hyper-box expanding condition.The experiment results indicate that the structure of fuzzy network constructed by this method is simpler and its performance is more efficient for pattern classification. |
| |
| Keywords: | neural network fuzzy neural network pattern recognition |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
|
