| 人工导向自组织子网学习方法研究 |
| |
| 引用本文： | 胡恒章 张贞子. 人工导向自组织子网学习方法研究[J]. 宇航学报, 1996, 17(4): 97-99 |
| |
| 作者姓名： | 胡恒章 张贞子 |
| |
| 作者单位： | 哈尔滨工业大学控制工程系 |
| |
| 摘 要： | 本文提出的人工导向自组织子网学方法，克服了子网结构中各子网间没有相互联系，对类别之间含混特征无法区分的缺点。通过在话者辨认实验中的应用表明，这种方法适于区分特征相近的类别，可用于辨别真假目标
|
| 关 键 词： | 自组织特征映射 子网 人工导向 |
USING FINE TUNING IN KOHONEN SUBNET STRUCTURE |
| |
| Hu Heng Zhang Zhang Zhen Zi. USING FINE TUNING IN KOHONEN SUBNET STRUCTURE[J]. Journal of Astronautics, 1996, 17(4): 97-99 |
| |
| Authors: | Hu Heng Zhang Zhang Zhen Zi |
| |
| Affiliation: | Department of Control Engineering Harbin Institute of Technology |
| |
| Abstract: | For using NN in time consuming identification task, it is a common way to break it into smaller subtasks according to the number of objects and to use a NN for each of the subtasks. But there are no relations among the subnets. So this structure can not distinguish objects with similar features. By using fine tuning in Kohonen subnet structure, such short-coming can be overcome. An experiment about speaker identification is carried out to show its effect. |
| |
| Keywords: | Self-organizing feature map Subnet Fine Tuning |
| 本文献已被 CNKI 维普 等数据库收录！ |
