Based On Discernibility Matrix Attribute Reduction Algorithm Research

Attributes reduction is one of important contents in rough set theory. Most traditional static attribute reduction algorithms can only handle static data sets, but the data in the real world is dynamic. Therefore, research on incremental reduction methods has important theoretical and practical value. Discernibility matrix is one kind of effective ways of attribute reduction. In this paper, incremental algorithms of attribute reduction based on discernibility matrix are studied by using information vector to reconstruct the discernibility matrix. The main research work can be summarized as follows:(1) A heuristic algorithm of attribute reduction based on discernibility matrix is presented. First, the positive and negative information vector and the discernibility matrix based on information vector are redefined. Second, the algorithms of constructing positive and negative information vector and constructing information vector matrix are given, and an attribute reduction algorithm is presented by using of frequency of attribute as heuristic condition. In the end, experiments validate the correctness and efficiency of the algorithm by using star spectra data.(2) An incremental algorithm of attribute reduction based on the discernibility matrix is presentd. Based on the above algorithm, an incremental attribute reduction algorithm is presented by analyzing the positive and negative domain, the information vector for dynamic data sets. Experiments validate the correctness and efficiency of the algorithm by using star spectra data.