Interactive fuzzy knowledge distance-guided attribute reduction with three-way accelerator

Attribute reduction with an accelerator is an efficient strategy to handle large-scale information systems, but it is rarely suitable for information systems with fuzzy decision (ISFD). Meanwhile, both three-way approximations (TA) and fuzzy knowledge distance (FKD) are two types of attractive uncertainty measures in ISFD. Therefore, it prompts us to deliberate on how to interact the merits of TA with FKD to develop an attribute reduction with a three-way accelerator for ISFD. Driven by this concern, the TA of fuzzy decision is used as prior knowledge to interact with FKD, then a model named ITA-FKD is constructed to explore the monotonicity of uncertainty in multi-granularity spaces and measure the significance of every attribute. A three-way accelerator is implemented to eradicate redundant objects and attributes with the aid of the ITA-FKD model. Subsequently, a filter-wrapper approximate attribute reduction with the unique accelerator is successfully devised. Ultimately, some comprehensive experiments with high-dimensional datasets containing biomedicine field exemplified the higher efficiency of the explored algorithm.