Método híbrido com detecção de regiões promissoras baseado em densidade para o problema de localização de rótulos cartográficos

• Ano de defesa: 2016
• Autor(a) principal: Araujo, Eliseu Junio [UNIFESP]
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• dARK ID: ark:/48912/001300002swhq
• Idioma: por
• Instituição de defesa: Universidade Federal de São Paulo (UNIFESP)
• Palavras-chave em Português: metaheuristcs; optimization; clustering search; metaheurísticas; otimização
• Link de acesso: https://sucupira.capes.gov.br/sucupira/public/consultas/coleta/trabalhoConclusao/viewTrabalhoConclusao.jsf?popup=true&id_trabalho=2907546; https://repositorio.unifesp.br/handle/11600/47285

Metaheuristcs have been the subject of research with the aim to find those having greater efficiency for solving optimization problems. It was noted during this operation, the hybrid metaheuristics are a good choice to accentuate the qualities of these methods. This project is focused on hybrid method Clustering Search (CS), focusing on the improvement and development of a new alternative for him, trying to make it an efficient, robust and flexible method in terms of quality solutions as well as computational time. CS seeks to combine heuristics and meta-heuristics for local search, intensifying the search for regions of space solutions considered promising. In this project we propose a new way to detect promising regions, based on clustering techniques DBSCAN, Label-propagation and NGI. To analyze this approach is proposed to solve a combinatorial optimization problem with many practical applications, the problem of location of map labels. In computational tests are used test problems from the literature. The results were satisfactory for Label-clusters made with propagation and NGI, showing better results than the CS, and showing that they are a good alternative to changing the method.