Incremental Approximation Feature Selection with Accelerator for Rough Fuzzy Sets by Knowledge Distance

Deyou Xia,Guoyin Wang,Qinghua Zhang,Jie Yang,Shuai Li,Man Gao
DOI: https://doi.org/10.1109/tfuzz.2023.3272157
IF: 12.253
2023-01-01
IEEE Transactions on Fuzzy Systems
Abstract:Feature selection method with rough sets based on incremental learning has the major advantage of the higher efficiency in a dynamic information system, which has attracted extensive research. However, the incremental approximation feature selection with an accelerator (IAFSA) remains ambiguous for a dynamic information system with fuzzy decisions (ISFD). Driven by this concern, the nonincremental approximation feature selection is first presented by fuzzy knowledge distance (FKD). Second, the incremental theory of FKD is constructed with a batch of objects appended to or removed from the dynamic ISFD. Subsequently, an acceleration mechanism to eliminate redundant information granules is developed to reduce the sample space. Eventually, two categories of IAFSA based on FKD are presented. The experiments reflect the efficiency and effectiveness of the developed IAFSA algorithms.
What problem does this paper attempt to address?