Population Diversity Management of Swallow Swarm Optimization Algorithm for Fuzzy Classification Problem

Abstract

In swarm algorithms, the need to measure population diversity arises in various contexts, such as in the adaptation of algorithm parameters, preventing the premature convergence of the algorithm and stopping and restarting it. Measures of population diversity allow the phases of the algorithm, namely, diversification and intensification, to be controlled. The article experimentally investigated six measures of population diversity of the optimization of the swallow swarm algorithm when solving the problem of optimizing the parameters of the membership functions of fuzzy classifiers. The resulting classifiers were tested on publicly available data sets drawn from the KEEL repository.