Robustness analysis of parameter settings in a genetic algorithm for the multi-objective shortest path problem

Abstract

The paper presents a sensitivity analysis of a genetic algorithm (GA) as applied to the multiobjective shortest path problem (MSPP). The study explores the robustness of the multi-objective evolutionary algorithm in terms of the sensitivity of its output to changes in its input parameters. Results show that mutation and the recombination operators only slightly affect the exploration and exploitation of solutions in the multiobjective shortest path problem. However, the size of the population and the number of generations of the GA largely influence the output of the genetic algorithm.