An incremental approach for updating approximations of rough fuzzy set under the variation of attribute values

Abstract

Rough Set Theory (RST) is a powerful mathematical tool for dealing with inconsistent information in decision situations. In real-life applications, information systems in RST often vary with time. Approximations of a concept in RST have been used to induce rules and need to update for dynamic data mining and related tasks. In addition, the values of the decision attributes in information systems may be fuzzy. An extension of classical rough set model, rough fuzzy set, is then presented to deal with such values. This paper focuses on approaches for dynamically updating approximations in rough fuzzy set when attribute values are coarsened or refined. The principles for dynamic maintenance of upper and lower approximations are firstly presented. Then, the algorithms are developed for updating approximations incrementally under the variation of attributes' values. Some examples are employed to illustrate the proposed methods. A comparison of the proposed incremental method with a non-incremental method for dynamic maintenance of approximations is conducted by an extensive experimental evaluation on the data set from UCI. The experimental results show that the incremental method effectively reduce the computing time in comparison with the non-incremental method.