Neuro-immune approach to solve routing problems
Tipo
Artigo
Data de publicação
2009
Periódico
Neurocomputing
Citações (Scopus)
17
Autores
Masutti T.A.S.
de Castro L.N.
de Castro L.N.
Orientador
Título da Revista
ISSN da Revista
Título de Volume
Membros da banca
Programa
Resumo
Routing problems constitute an important class of problems because they are hard to solve and have a wide variety of real-world applications. Two of the most well-studied routing problems are the traveling salesman problem (TSP) and the multiple traveling salesmen problem (MTSP). This paper proposes the use of a neuro-immune self-organizing network to solve both, TSP and MTSP. The algorithms are introduced and a brief review of the literature on self-organizing networks, applied separately to each of these problems, is performed. To assess the performance of the proposed algorithms, a number of tests are conducted on benchmark data, and the results are directly compared with those of similar works. The results obtained show a better performance in terms of the best solution found for many instances, and an equivalent average performance. The main advantages of the proposed approach are its network constructive architecture and the use of a single growing network to solve the MTSP. © 2009 Elsevier B.V. All rights reserved.
Descrição
Palavras-chave
Assuntos Scopus
Artificial immune systems , Artificial neural networks , Benchmark datum , Multiple traveling salesmen problem , Real-world applications , Routing problems , Self-organizing networks