| 一种高效的自组织特征映射图的初始化方法 |
| |
| 引用本文： | 张群洪,陈崇成. 一种高效的自组织特征映射图的初始化方法[J]. 计算机科学, 2008, 35(3): 128-131 |
| |
| 作者姓名： | 张群洪 陈崇成 |
| |
| 作者单位： | 福建省空间信息工程研究中心,福州大学数据挖掘与信息共享教育部重点实验室,福州,350002 |
| |
| 基金项目： | 国家自然科学基金 , 福建省科技计划 |
| |
| 摘 要： | 自组织特征映射图算法(SOFM,self-organizing Feature Map)在模式识别中有着广泛的应用.本文首先讨论了网络结构的初始化设置对自组织特征映射图构造的影响以及加速SOFM网络学习训练过程的主要方法,然后提出一种从边界到中心的自组织特征映射图初始化方法,该方法形成的自组织特征映射图能够真实地表示输入样本内在关系,大大减少学习训练次数,从而有效改进了传统的SOFM算法.
|
| 关 键 词： | 自组织特征映射图 初始化方法 加快SOFM学习过程 |
An Efficient Initialization Scheme for Self-organizing Feature Map |
| |
| ZHANG Qun-Hong,CHEN Chong-Cheng. An Efficient Initialization Scheme for Self-organizing Feature Map[J]. Computer Science, 2008, 35(3): 128-131 |
| |
| Authors: | ZHANG Qun-Hong CHEN Chong-Cheng |
| |
| Affiliation: | ZHANG Qun-Hong CHEN Chong-Cheng(Spatial Information Research Center of Fujian,Key Lab of Data Mining & Information Sharing of Ministry of Education,Fuzhou University,Fuzhou 350002) |
| |
| Abstract: | Self-Organizing Feature Map algorithm (SOFM) is widely applied in various areas such as pattern recognition. The paper first discusses the influence of initialization on self-organizing feature map and the main methods to accelerate learning phase of self-organizing feature map. Then an efficient initialization scheme for forming a topologically ordered feature map from boundary to center is proposed. The scheme can enormously reduce the training epochs of self-organizing feature map which can present the t... |
| |
| Keywords: | Self-organizing feature map Initialization scheme Accelerate learning phase |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
点击此处可从《计算机科学》下载全文 |
