Diversity study of multi-objective genetic algorithm based on Shannon entropy

Multi-objective optimization inspired on genetic algorithms are population based search methods. The population elements, chromosomes, evolve using inheritance, mutation, selection and crossover mechanisms. The aim of these algorithms is to obtain a representative non-dominated Pareto front from a given problem. Several approaches to study the convergence and performance of algorithm variants have been proposed, particularly by accessing the final population. In this work, a novel approach by analyzing multi-objective algorithm dynamics during the algorithm execution is considered. The results indicate that Shannon entropy can be used as an algorithm indicator of diversity and convergence.