Rough sets-based machine learning using a binary discernibility matrix
R. Felix, Toshimitsu Ushio · 1999
This paper presents an approach with two methods to obtain minimal coverings in rough sets based machine learning, both methods are based on the definition of a binary discernibility matrix. The first method is an exhaustive search of coverings and the second uses a genetic algorithm (GA) based search. The approach represents the discernibility of two examples by a condition attribute of an information system in a single bit. Thus, operations that usually are performed with a set approach are redefined in order to use bit-wise logical operations. The algorithms for both methods are presented and discussed.