During last years, a wide range of population-based metaheuristics, such as evolutionary algorithms and swarm-based approaches, among others, have been proposed with the aim of dealing not only with benchmark optimization problems, but also with real-world applications belonging to a significant number of fields. Population-based approaches maintain a set of solutions with the aim of exploring the search space in an efficient way. Usually, a diverse set of solutions is maintained, meaning that several regions are explored simultaneously. However, one common problem of population-based metaheuristics is that for some test cases they might exhibit a tendency to converge quickly towards some regions. In fact, one of the most frequent drawbacks that these types of metaheuristics have to deal with is premature convergence, which arises when every member of the population is located at a sub-optimal area of the decision space from where they cannot escape. Differently, in some cases the population might diverge and promising regions located during the search might not be properly exploited. Many practitioners consider the proper management of diversity as one of the cornerstones for proper performance. As a result, a significant number of methods to implicitly or explicitly manage the diversity in a set of solutions have been proposed, e.g., mating-based approaches, disruptive operators, fitness sharing, crowding-based selection, and methods based on complex population structures, among others. Furthermore, some more recent proposals such as diversity-based multi-objective approaches and multi-objectivization, have gained popularity during last years in the case of promoting diversity by means of the application of multi-objective algorithms to single-objective problems. Since some of the methods that are used in multi-objective optimization to preserve diversity in the objective space are similar to those used in single-objective optimization in the decision space, some advances in one of the fields might guide the advances in the other one.

The fourth edition of this special session aims to attract studies related to diversity in population-based metaheuristics. The specific topics covered in this special session are detailed in the "Topics" section.