摘要
模式是事物的本质,特征只是事物的表象,模式识别的任务就是要建立事物本质和表象之间的联系,传统模式识别是通过分类器建立这样的联系,模式识别的效率依赖于特征提取和选择。该文基于信息增益,通过动力学系统演化方程建立事物本质和表象之间的联系,该过程更符合人的认知过程。
The pattern is nature of matter,and character is only appearance.The task of pattern recognition is to construct concern between nature and appearance.Traditional pattern recognition does this by classifier,and the efficiency of pattern recognition depends on character extraction and choosing.This paper constructs concern between nature and appearance by dynamic system evolution equation based on information augmentation,that accords with human's learning procedure.
出处
《计算机工程与应用》
CSCD
北大核心
2005年第15期51-52,190,共3页
Computer Engineering and Applications
基金
江苏省博士创新基金资助(编号:XM-35)
关键词
模式识别
信息增益
动力学系统演化
pattern recognition,information augmentation,evolution equation of dynamic system
