Evolving Fuzzy Reasoning Approach Using a Novel Nature-Inspired Optimization Tool

Abstract In general, fuzzy reasoning tool with Mamdani approach has good readability, but low accuracy; whereas the same with Takagi and Sugeno’s approach ensures high accuracy but at the cost of readability. In the developed combined form of fuzzy reasoning, the merits of both the above two approaches are utilized to obtain both the high accuracy as well as good readability. The above combined form is evolved using a recently-developed nature-inspired technique, namely Bonobo Optimizer (BO). This optimization method mimics the fission-fusion social structure and reproductive schemes adopted by bonobos. In addition, controlling parameters of the BO are designed to be adaptive and self-adjusting to perform efficiently for a variety of problems. The performances of the developed models have been tested and compared with that of the combined fuzzy reasoning tool evolved using Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Grey-Wolf Optimizer (GWO) and Jaya Algorithm for three data sets. The novelty of this study lies with the ability of the recently proposed BO to evolve an efficient fuzzy reasoning approach.