An Accelerator for Rule Induction in Fuzzy Rough Theory

Rule-based classifier, that extract a subset of induced rules to efficiently learn/mine while preserving the discernibility information, plays crucial role in human-explainable artificial intelligence. However, this era big data, rule induction on whole datasets is computationally intensive. So far, best our knowledge, no known method focusing accelerating has been reported. This first study consider acceleration technique reduce scale computation induction. We propose an accelerator for based fuzzy rough theory; can avoid redundant and accelerate building classifier. First, consistence degree, called consistence-based value reduction (CVR), proposed used as basis accelerate. Second, we introduce compacted search space termed key set, which only contains instances required update rule, conduct reduction. The monotonicity set ensures feasibility accelerator. Third, rule-induction designed it theoretically guaranteed display same results unaccelerated version. Specifically, rank preservation property consistency between achieved by method. Finally, extensive experiments demonstrate perform remarkably faster than rule-based classifier methods, especially with numerous instances.