属性约简准则与约简信息损失的研究

发布时间：2019-07-27 08:23

【摘要】：属性约简是粗糙集的重要研究内容,信息熵是度量信息量的方法.在研究绝对约简和几种相对约简的基础上,归纳出属性约简的一般准则.定义了基于条件属性信息熵的属性约简和基于联合熵的属性约简,研究了几种属性约简与绝对约简之间的关系.定义了基于条件属性信息熵的约简信息损失,澄清了属性约简不损失信息的含糊观念,指出了属性约简只是在约简准则意义下不损失信息,在信息熵意义下可能损失信息.为进一步研究粗糙集、粒计算中属性约简与分类夯实了信息论基础.
[Abstract]:Attribute reduction is an important research content of rough set, and information entropy is a method to measure information. Based on the study of absolute reduction and several relative reduction, the general criteria of attribute reduction are summarized. Attribute reduction based on conditional attribute information entropy and attribute reduction based on joint entropy are defined, and the relationship between several attribute reduction and absolute reduction is studied. This paper defines the reduction information loss based on conditional attribute information entropy, clarifies the vague concept that attribute reduction does not lose information, and points out that attribute reduction only does not lose information in the sense of reduction criterion, and may lose information in the sense of information entropy. In order to further study rough sets, attribute reduction and classification in grain computing consolidate the basis of information theory.
【作者单位】：浙江师范大学数理与信息工程学院;浙江师范大学行知学院;同济大学电子与信息工程学院;
【基金】：国家自然科学基金(No.61572442,No.61203247,No.61273304,No.61573259,No.61472166) 浙江省自然科学基金(No.LY15F020012) 浙江省自然科学青年基金(No.Q13F020006)
【分类号】：TP18

【相似文献】