关键词:

模糊粗糙集, 属性约简, 贴进度, 可辨识矩阵

Abstract:

Attribute reduction is the main research field of fuzzy rough sets， also the first systematic application of that. However， some scholars have proved that the problem of NP-hard is to find the minimum reduction. In this paper， a new reduction algorithm is proposed， in which the discernible matrix is first generated by the similarity degree， and the relative core attributes are obtained by the discernible matrix， then the heuristic algorithm based on the core attributes to generate the minimum reduction is presented. Finally， an example is given to illustrate the effectiveness of the method.

Key words: fuzzy rough sets, attribute reduction, similarity degree, discernible matrix