Incremental updating fuzzy rough approximations for dynamic hybrid data under the variation of attribute values

With the development of the Internet of Things (IoT), Hybrid Information Systems (HIS) collect increasing number of hybrid data. A novel Gaussian kernel Fuzzy Rough Sets (FRS) was constructed based on a new hybrid distance in our previous study. In real applications, with the deepening of cognition or improvement of technology, attribute values often change. There are three cases of changes: missing values are imputed, error values are corrected and values are coarsened or refined. In this paper, the mechanisms of attribute values changes and fuzzy e-quivalence relation in FRS are analyzed, and several incremental approaches for updating approximations are discussed.