Multi-Objective Animal Migration Optimization - A Metaheuristic Optimization Algorithm

Abstract

Metaheuristic optimization has received a lot of attention over the last couple of decades. Some optimization problems are just too computationally heavy to solve with traditional search techniques, especially when there is more than one objective to optimize for. There have been many promising metaheuristic algorithms for solving multi-objective problems, but algorithms have different strengths and weaknesses, and no algorithm can be the best at solving every problem. New metaheuristic approaches are, therefore, an interesting topic to study. In this thesis, a structured literature review of state of the art single and multi-objective metaheuristic algorithms was performed. A framework was created to implement and evaluate promising algorithms. Moreover, a comparison study of single-objective algorithms was performed, where the most suitable single-objective algorithm was extended to handle multi-objective problems. The extended algorithm's performance was compared with other well-performing algorithms from the literature. The resulting algorithm is called Multi-Objective Animal Migration Algorithm (MOAMO) and showed competitive results when compared with eight other algorithms on 22 test functions using three performance metrics.