I. Introduction
Attribute reduction is one of the most important subjects of Rough Set theory (RST), which is considered as a part of the most useful methods for data mining and knowledge discovery. Attribute reduction is a process to find optimal subset from system to effectively represent the giving dataset. According to the complexity of real life data, finding all minimal attribute reduction considered as a NP-Hard problem. Over the past years, researches proposed many heuristic and meta-heuristic algorithms to find optimal solution in attribute reduction. For instance, ant colony, genetic algorithm and simulated annealing algorithms by Jensen and Shen [3], [5] and [9], scatter search by Wang et al. [7], and tabu search by Hedar et al. [8]. Further reading on other approaches and surveys on rough set attribute reduction can be found in [10], [11].