|
In this paper, a zero-one discernibility matrix to measure the classification ability of each attribute is proposed, and the properties of binary matrix are discussed. The relationship between the attribute reduction and the binary matrix covering is investigated theoretically. The problem of searching an attribute reduct is transformed into the problem of searching a set of binary matrices, in which the union of all binary matrices may cover the objective binary matrix. A heuristic reduction based on binary matrix is presented. For an information table, each attribute is associated with a measure based on binary matrix to denote the significance of the attribute. In the proposed heuristic reduction algorithm, attribute reduct is constructed by adding attributes in the sequence of attribute significance. The complexity of the algorithm is analyzed, and the validity of the e algorithm is tested based on an illustrative example. |
|
Keywords:attribute reduction; rough set; descernibility matrix; zero-one matrix |
|