Scalable Fuzzy Genetic Classifier Based on Fitness Approximation

Fuzzy classification rules are widely used for classification, as they are more interpretable as well as efficient in handling the real-world problems, which involves imprecision and vagueness. Genetic algorithms are proven stochastic search techniques employed in automatic generation of fuzzy classification rule. However, genetic algorithms employed for the said task require large number of fitness evaluation or performance evaluations in achieving a reasonable solution requiring a large amount of computational time. Hence, to expedite the execution is a major concern in genetic algorithms. In this paper, we incorporate fitness inheritance mechanism in genetic algorithms to design a scalable genetic fuzzy classifier, which reduce the number of actual fitness function evaluations of subsequent generations and produce rules with acceptable classification accuracy.