A rough sets & genetic based approach for rule induction

Automated knowledge acquisition is an important research area in developing expert systems. For this purpose, several methods of inductive learning have been proposed, such as decision tree, fuzzy set, Dempster-Shafer theory of evidence. However, most of the approaches require prior knowledge. Rough sets theory is a new approach to decision making in the presence of uncertainty and vagueness, when coupled with genetic algorithms, a rule induction engine that is able to induce rules efficiently. In this paper, we use discernibility matrix to find the attribute core. Then a steady-state genetic algorithms is applied to get relative reduction, finally, rough sets theory is used to remove redundant condition attribute values to get rules. The experimental results show that the proposed method induces maximal generalized rules efficiently.