Novel fuzzy β-covering rough set models and their applications

In recent years, fuzzy β-covering, as a natural extension of fuzzy coverings, has attracted considerable attention. However, existing fuzzy β-neighborhood operators cannot accurately describe the relationship between objects, which greatly restricts the application of fuzzy β-covering. For this reason, we first construct four new fuzzy β-neighborhood operators by using the existing fuzzy β-neighborhood operator and generalized fuzzy logic operators, and investigate their properties. To better portray the similarity between samples, inspired by the definition of fuzzy similarity relation, we define the concept of fuzzy β-covering relation. On this basis, we develop a new framework of fuzzy β-covering rough set models. We further propose an attribute reduction method by employing the new fuzzy β-covering relation, and design a heuristic attribute reduction algorithm with reference to an uncertainty measure called attribute significance. Finally, experimental results show the superiority of our proposed method through a series of experimental analyses.