Improved algorithm for attribute core computing based on binary discernibility matrix

摘要

To solve the irrational problem of conventional algorithms for attribute core computing, an improved binary discernibility matrix based algorithm for attribute core computing is proposed. A threshold for determining which one is the core attribute is introduced into the algorithm for adjusting the inflexible determinant conditions. For initial obtained core attributes by traditional binary discernibility matrix based algorithm, a uniform threshold is used to filter out those attributes with small probability to become solid core attributes. The simulation results show that the improved algorithm has a good performance, can improve the rationality and accuracy of subsequent attribute reduction.