Preface to the Special Issue on Developments in Metaheuristics

This special issue includes mainly, but not only, papers presented at the 13th Metaheuristics International Conference (MIC 2019) held in Cartagena (Colombia) from July 28 to 31, 2019. This edition of MIC was the perfect setting for attendees to disseminate and keep up to date with the progress in metaheuristics, their techniques, empirical and theoretical research, industrial applications, and interface with other domains. The conference brought 100 participants (37 of them students) from 17 countries. Bringing this community to the Caribbean was a special and magical experience. The plenaries, delivered by leaders of the optimization community, were the opportunity to explore the past and the present of the field, and to look at promising future opportunities. They included a historical perspective of the field illustrated with several stories behind the three editions of the Handbook of Metaheuristics and the discussion of potential challenges and opportunities that arise at the interface between machine learning and metaheuristics. Likewise, the tutorials and plenaries included fresh views and applications of metaheuristics to problems tackled regularly by the optimization community (routing and scheduling application in an urban context, as well as packing problems and their intersection with routing that naturally arise in industrial applications). They also included novel applications of population metaheuristics to generate realistic scenarios of stochastic optimization problems and the use of metaheuristics for the solution of polarization problems that we face in our society at a global or local level. After a careful peer-review process, this special issue features eighteen articles covering several metaheuristic applications. Routing and scheduling problems that have been a classical application domain of metaheuristics continue to receive an important attention of the research community. This special issue is not an exception, and twelve out of the eighteen articles deal with such problems. Remarkably, the use of powerful hybrid metaheuristics allows the inclusion of many realistic features in the modeling of these problems. Current environmental and social sustainability issues lead to new features in the classical models or give rise to new optimization problems and application domains (e.g., energy consumption and labor regulation in routing problems and districting decisions with max dispersion criteria for the collection of waste of electric and electronic equipment). Likewise, classical scheduling problems (e.g., permutation flow shop scheduling) are used to test new strategies in the automated design of metaheuristic and some others are enhanced including multi-skills and transportation stages. Nevertheless, other less studied application domains (time-series forecasting, telecommunications, information retrieval, and humanitarian logistics) and techniques (robust and bilevel optimization) are also present in this special issue. Finally, real-world applications inspired by internal logistics, rural vehicle routing, and long-haul transportation are also addressed with metaheuristics presented in this issue. We believe this special issue brings the reader a broad and deep picture of the current state-of-the-art of metaheuristics for the solution of complex (combinatorial) optimization problems and the many available options in the design and implementation of these methods that do not rely on complex and artificial (and unnecessary) metaphors. Solution methods in this special issue include: (i) population metaheuristics featuring novel designs, like multi-start biased-random-key genetic algorithms (BRKGAs) and memetic algorithms with multiple crossover operators; (ii) hybrid metaheuristics combining genetic algorithms, greedy randomized adaptive search procedures (GRASP), iterated local search (ILS), tabu search, variable neighborhood search (VNS), large neighborhood search (LNS), record-to-record travel, infeasible solution acceptance, strategic oscillation and repair strategies, multi-start methods, and path relinking, among others; (iii) matheuristics combining metaheuristics and mathematical programming models to produce initial or partial solutions for a metaheuristic or to guide the search and explore large neighborhoods, (iv) machine learning techniques trained with problem features and algorithm performance to recommend the best metaheuristic and its parameter settings to solve a previously unseen problem; and, finally, (v) simheuristics that integrate simulation models and metaheuristics for the solution of combinatorial optimization problems with stochastic parameters (e.g., SimILS). It is noteworthy that the papers on this special issue follow suitable practices for the design, fine-tuning, and evaluation of the proposed metaheuristics. This includes the use of automatic or statistically supported fine-tuning procedures (eg., the irace package); the use of time-to-target plots and non-parametric statistics to compare the performance of several algorithms, and the separated evaluation of components to assess their contribution to the overall performance of the proposed hybrid metaheuristics.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.