Dynamic tabu search for dimensionality reduction in rough set

This paper proposed a dynamic tabu search (DTSAR) that incorporated a dynamic tabu list to solve an attribute reduction problem in rough set theory. The dynamic tabu list is use to skip the aspiration criteria and to promote faster running times. A number of experiments have been conducted to evalute the performance of the proposed technique with other published metaheuristic techniques, rough sets and decision tree. DTSAR shown promising results on reduct generation time. It ranges between 0.20 minutes to 22.18 minutes. For comparison on the performance on number of reduct produced, DTSAR is on par with other metaheuristic techniques. DTSAR outperforms some techniques on certain dataset. Quality of classification rules generated by adopting DTSAR was comparable with two other methods i.e. Rough Set and Decision Trees.