| 基于ART1改进算法的汉字分类研究 |
| |
| 引用本文： | 代小娟,林小竹,周小正.基于ART1改进算法的汉字分类研究[J].北京石油化工学院学报,2009,17(1):38-42. |
| |
| 作者姓名： | 代小娟 林小竹 周小正 |
| |
| 作者单位： | 1. 北京石油化工学院自动化系,北京,102617;北京化工大学信息技术学院,北京,100029
2. 北京石油化工学院自动化系,北京,102617 |
| |
| 摘 要： | 为了更好地利用自适应共振理论（ART1）来实现汉字的分类，提出了一种改进的ART1算法。改进算法引用了同或的思想，将输入模式与记忆模式相同的部分在输入模式总体中占的比例作为二者的匹配度，从而降低了输入样本顺序对分类结果造成的误差，增强了网络的稳定性和抗噪性。实验结果证明，改进算法能够更好地将汉字进行分类，求出聚类中心。
|
| 关 键 词： | 神经网络 自适应共振理论 警戒参数 聚类中心 |
Chinese Classification Based on Improved ART1 Algorithm |
| |
| Dai Xiaojuan,Lin Xiaozhu,Zhou Xiaozheng.Chinese Classification Based on Improved ART1 Algorithm[J].Journal of Beijing Institute of Petro-Chemical Technology,2009,17(1):38-42. |
| |
| Authors: | Dai Xiaojuan Lin Xiaozhu Zhou Xiaozheng |
| |
| Affiliation: | Dai Xiaojuan Lin Xiaozhu Zhou Xiaozheng (1 Department of Automation, Beijing Institute of Petro-chemical Technology, Beijing 102617; 2 School of information technology, Beijing University of Chemical Technology, Beijing 100029) |
| |
| Abstract: | In order to make better use of adaptive resonance theory （ART1） to achieve the classification of Chinese characters, an improved algorithm of ART1 was proposed. The algorithm quotes the thought of XNOR, the percentage of the same part between the input and the memory mode in the overall input mode was used to define the match degree, thus the classification errors of the input sample sequence is reduced and the stability and the noise immunity of the network are enhanced. The experimental results proved that the improved algorithm can classify Chinese characters better and the cluster center can be calculated. |
| |
| Keywords: | artificial neural network adaptive resonance theory vigilance parameter clustercenter |
| 本文献已被 维普 万方数据 等数据库收录！ |
|
