A metaheuristic with self-construction of operators for global continuous optimization: extensions and applications of the drone squadron optimization algorithm

• Grant number: 17/20844-0
• Support Opportunities: Regular Research Grants
• Start date: March 01, 2018
• End date: November 30, 2018
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computer Systems
• Principal Investigator: Vinícius Veloso de Melo
• Grantee: Vinícius Veloso de Melo
• Host Institution: Instituto de Ciência e Tecnologia (ICT). Universidade Federal de São Paulo (UNIFESP). Campus São José dos Campos. São José dos Campos , SP, Brazil
• Associated researchers: Alexandre Cláudio Botazzo Delbem

Abstract

Several meta-heuristics have been proposed for solving various types of optimization problems. In general, they use solution modifying procedures that combine current solutions to generate new solutions and find increasingly better solutions over the iterations. To improve performance, one can develop more efficient operators or add strategies for meta-heuristics to be adaptive, automatically changing their behavior during the optimization process. This development requires creativity and background knowledge from the researcher to identify some information that can improve the performance of the meta-heuristic. Later, such information must be inserted into the meta-heuristic in some way that is effective. With hyper-heuristic techniques, it is possible to automate part of the creation of algorithms, taking from the researcher the task of finding this improvement through trial and error. In this sense, a hybrid technique was created, called Drone Squadron Optimization, which uses a hyper-heuristic to create the modifying procedures during the optimization process performed by the meta-heuristic (in an online approach), so that it becomes self-adaptive. This approach differs from others in the literature, which in general work offline (build a procedure and evaluate its performance in a complete meta-heuristic execution) or select from pre-coded procedures. In this project, the proponent and collaborators (among them, undergraduates and Professors) will extend the technique to other lines of research (memetic optimization and constrained optimization) and apply them in several other optimization problems in the literature. (AU)

Scientific publications

(References retrieved automatically from Web of Science and SciELO through information on FAPESP grants and their corresponding numbers as mentioned in the publications by the authors)

DE MELO, VINICIUS V.; LORENA, ANA C.; IEEE. Using Complexity Measures to Evolve Synthetic Classification Datasets. 2018 INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS (IJCNN), v. N/A, p. 8-pg., 2018-01-01. (17/20844-0, 12/22608-8)