新的信息熵属性约简

New Attribute Reduction on Information Entropy

在粗糙集不确定性度量公式中,模糊熵和模糊度是重要的度量方式。根据粗糙集不确定性度量中模糊熵和新的模糊度公式,提出了在决策信息系统中修正条件信息熵和相对模糊熵的概念,并分别用两种方式证明了熵在属性约简过程中的单调性。然后利用向前添加属性算法进行属性约简,约简结果在RIDAS(roughset based intelligent data analysis system)平台上进行识别率测试,通过实验对比分析了两种新的信息熵与条件信息熵的约简结果,为基于信息熵的属性约简提供了参考。

Fuzzy entropy and fuzziness are both important measurement methods of rough set uncertain measure. According to the formula of fuzzy entropy and new fuzziness of rough set uncertain measure, this paper proposes the concepts of revisionary conditional information entropy and relative fuzzy entropy in the decision information system, and proves the monotonicity of entropy by using two methods respectively in the process of attribute reduction. Then, this paper uses the algorithm of adding attributes for attribute reduction, and tests the recognition rate in RIDAS （rough set based intelligent data analysis system） platform. Finally, this paper compares and analyzes the experimental results between the two proposed information entropy and conditional information entropy, and provides a reference for the attribute reduction based on information entropy.