On the Sensitivity of a Bee-Inspired Algorithm to Its Internal Parameters

Abstract

Over the past two decades a wide range of nature-inspired clustering algorithms has been proposed in the literature with competitive performance when applied to solving real-world complex problems. One common feature of most of these algorithms is the need to set a number of internal parameters so that they can be suitably applied to these problems. These parameters are usually introduced so as to simplify or model some biological aspects of the phenomenon being modeled, but they greatly influence the performance of the proposed algorithm. The present paper takes one of these bio-inspired algorithms and investigates how its input, user-defined, parameters influence its performance. By doing that, we provide some important clues and guidelines for potential users to understand how to better set the parameters so as to take the most out of the algorithm. Two different versions of the algorithm are considered in the analysis reported here.