A novel approach to obtain relative reduct

Knowledge reduction is an important issue when dealing with huge amounts of data. And it has been proved that computing the minimal reduction of decision system is NP-complete. This paper proposes a novel approach to obtain relative reduct of a decision system whose attributes values are expressed by fuzzy sets in rough set-based machine learning. Some properties of the discernibility relation matrix are presented first The binary discernibility relation is used to find relative reduct directly, based on these properties. A heuristic information is presented in the process of selecting the attribute of relative reduct The algorithm is presented and a simple example is detailed to illustrate the approach.