Current research and future perspectives on decision-theoretic rough sets

Abstract

As the central concepts in rough set theory, the classical Pawlak lower and upper approximations are defined based on qualitative set-inclusion and non-empty overlapping relations, respectively. Consequently, the theory suffers from an intolerance of errors, which greatly restricts its real-world applications. To overcome this limitation, Yao and colleagues proposed a decision-theoretic rough sets (DTRS) model in early 1990s' by introducing the Bayesian decision theory into rough sets. In recent years, the model has attracted much attention and has been applied in uncertain information processing. This paper aims at (1) presenting a survey of the motivations for introducing the DTRS model, the main features of the model, and the problems to be studied in the model, (2) reviewing the fundamental results, state-of-art research, and challenges, and (3) pointing out future perspectives and potential research topics. ©, 2015, Science Press. All right reserved.