Different meta-heuristics (MHs) may find the best solutions for different traveling salesman problem (TSP) instances. The a priori selection of the best MH for a given instance is a difficult task. We address this task by using a meta-learning based approach, which ranks different MHs according to their expected performance. Our approach uses Multilayer Perceptrons (MLPs) for label ranking. It is tested on two different TSP scenarios, namely: re-visiting customers and visiting prospects. The experimental results show that: 1) MLPs can accurately predict MH rankings for TSP, 2) better TSP solutions can be obtained from a label ranking compared to multilabel classification approach, and 3) it is important to consider different TSP application scenarios when using meta-learning for MH selection. © 2012 Springer-Verlag.
CITATION STYLE
Kanda, J., Soares, C., Hruschka, E., & De Carvalho, A. (2012). A meta-learning approach to select meta-heuristics for the traveling salesman problem using MLP-based label ranking. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 7665 LNCS, pp. 488–495). https://doi.org/10.1007/978-3-642-34487-9_59
Mendeley helps you to discover research relevant for your work.