I. Introduction
The increasing amount of data stored in real-world databases make the process of extracting useful knowledge from data a challenging and complex task. The challenges and complexity of extracting useful knowledge from data has motivated researchers to reduce the data dimensionality in order to make the data mining process more efficient. Therefore, attribute reduction is an active research area in data mining filed. The main aim of attribute reduction is to find a subset of attributes that are relevant for the target data mining task (Pawlak [1] [2]). Given a dataset with N attributes, the goal is to find the minimal reduct by removing irrelevant attributes which can improve the performance and efficiency of an applied learning process. The optimal reduct is assessed by both relevancy and redundancy aspects. Finding an optimal subset of attributes varies from one problem to another depending on the problem complexity and size. Furthermore, it is known that the process of determining a minimal reduct is a NP-hard problem Pawlak [1].