连续值信息系统的不确定性度量

Abstract

不确定性的度量方法是人工智能研究的重要课题之一，受到国内外专家学者的广泛关注，相关研究成果已经成功的应用于数据挖掘，决策分析，模式识别与人工智能领域中。通过二元关系与熵，对连续值信息系统中的不确定性度量进行了系统研究。基于经典Pawlak粗糙集理论中的近似精度、知识粒度与信息熵，提出了连续值信息系统的粗糙度、知识粒度与知识熵，并对三种度量方式进行了比较分析。三种不确定性度量方式的提出，为连续值信息系统知识约简与表示的研究提供了理论基础。 Approach to uncertainty measures is one of the hottest topics in area of artificial intelligence, which attracts attention from many researchers. Relevant research results have been applied in data mining, decision analysis, pattern recognition and artificial intelligence. In this paper, the uncertainty measures of continuous-valued information systems have been investigated system-atically by using binary relation and entropy. Based on the approximation accuracy, knowledge granulation and information entropy in Pawlak rough set theory, we propose the rough accuracy, knowledge granulation and knowledge entropy in continuous-valued information systems. We also have the comparative study about three measures in the paper. The proposed uncertainty measures for continuous-valued information systems could provide the theoretical foundation for knowledge reduction and representation in continuous information systems.