ABSTRACT

There are different types of techiniques are there to extract knowledge from various sources. Critical / rough set has been applied to extract knowledge from various types of databases. Some limitations have been discovered in rough set, such as label inconsistency, the lack of flexibility and excessive dependency on discretization of the initial attributes. To overcome these limitations, a novel agglomerative clustering method using improved rough set is proposed. The idea of using equivalence class was also incorporated to merge and divide subclass. The experimental applications in data extraction and cooperative object localization showed the effectiveness of the presented improved rough set combined with agglomerative clustering.