A Modified Symbiotic Organism Search Algorithm with Lévy Flight for Software Module Clustering Problem

To date, there are much increasing trends on adopting parameter free meta-heuristic algorithms for solving general optimization problems. With parameter free algorithms, there are no parameter controls for tuning. As such, the adoption of parameter free meta-heuristic algorithms is often straightforward. On the negative note, exploration (i.e. roaming the search space thoroughly) and exploitation (i.e. manipulating the current known best neighbor) are pre-set. As the search spaces are problem dependent, any pre-set exploration and exploitation can lead to entrapment in local optima. In this paper, we investigate the use of Lévy flight to enhance the exploration of a parameter free meta-heuristic algorithm, called Modified Symbiotic Organism Search Algorithm (MSOS), via its population initialization. Our experimentations involving the software module clustering problems have been encouraging, as MSOS gives competitive results against existing selected parameter free meta-heuristic algorithms. For all the given module clustering problems, MSOS generates overall best mean results.