Attribute selection for rough sets is an NP-hard problem, in which fast heuristic algorithms are needed to find reducts. In this project, two reduct methods for rough set were implemented: particle swarm optimization and Johnson’s method. Both algorithms were evaluated with five different benchmarks from the KEEL repository. The results obtained from both implementations were compared with results obtained by the ROSETTA software using the same benchmarks. The results show that the implementations achieve better correction rates than ROSETTA.
Li, Xiaohan, "Attribute Selection Methods in Rough Set Theory" (2014). Master's Projects. 352.