Detecting Overlapping Areas in Unbalanced High-Dimensional Data Using Neighborhood Rough Set and Genetic Programming