Attribute Reduction Algorithm Based on Genetic Algorithm

The most issue is designing the fitness function of the chromosome when Generic algorithm is been used for gcalculating the minimal attribute reduction in rough set theory. But with the existed fitness function of the chromosome, the one that the value of the fitness function is larger might not be an attribute reduction. So the optimization candidate attribute reduction might not be the minimal attribute reduction. What is more, during the crossover and mutation process, it could not delete the candidate attribute reduction which is not the minimal attribute reduction. To solve the mentioned problems and speed up the convergence speed. In this paper, a new fitness function is introduced, and proved that the optimization candidate attribute reduction must be an attribute reduction. It also can delete the candidate attribute reduction which is not the minimal attribute reduction in the crossover and mutation process. Then an efficient attribute reduction algorithm based on genetic algorithm is proposed. The results of experiment show that the new algorithm may find the minimal attribute a reduction and has quick convergence speed.