| 引用本文: | 虞水俊,孔铁生,梁甸农.基于神经网络的特征结构提取[J].国防科技大学学报,1995,(1):16-21.[点击复制] |
| Yu Shuijun,Kong Tiesheng,Liang Diannong.Neural-network Based for Eigenstructure Extraction[J].Journal of National University of Defense Technology,1995,(1):16-21[点击复制] |
|
| |
|
|
| 本文已被：浏览 1841次 下载 1514次 |
| 基于神经网络的特征结构提取 |
| 虞水俊, 孔铁生, 梁甸农 |
|
(国防科技大学 电子技术系 湖南 长沙 410073)
|
| 摘要: |
| 本文在详细讨论特征结构提取问题的能量函数表示的基础上，通过对该能量函数表示的适当变换和整理，高阶关联非线性连续时间神经网络被引入到特征结构的提取问题中。这种方法直观、明了，它将所要提取的特征结构与网络稳定时的输出直接对应起来，可对特征结构进行自适应跟踪、估计，这为特征结构的提取提出了一个新的方法。 |
| 关键词: 特征结构 能量函数 神经网络 |
| DOI： |
| 投稿日期：1994-03-21 |
| 基金项目: |
|
| Neural-network Based for Eigenstructure Extraction |
| Yu Shuijun, Kong Tiesheng, Liang Diannong |
|
(Department of Electronic Technology)
|
| Abstract: |
| In the paper，The expression of the cost function of the eigenstructure extraction problem is discussed in detail. Through a proper transformation of the expression of the cost funtion，high-order neural network is introduced into the eigenstructure extraction problem. According to the method，the eigenstructure to be extracted is connected with the output of the network directly when the network is stable. The method is simple and direct，and can be used to track the eigenstructure adaptively. It opens a new area for the extraction of the eigenstructure. |
| Key words: eigenstructure cost fuction neural network |
|
|
