论文部分内容阅读
本文提出一种用于大类别模式识别系统的改进的自组织聚类方法．采用这种方法把３７５５种多体印刷汉字聚类为几千个子集．聚类正确率优于９８％．
This paper presents an improved self-organizing clustering method for large-scale pattern recognition systems. Using this method, 3755 multi-body printed Chinese characters are clustered into thousands of subsets. The correct rate of clustering is better than 98%.