Local Optima Networks of the Permutation Flow-Shop Problem

Fabio Daolio (1) , Sébastien Verel (2) , Gabriela Ochoa (3) , Marco Tomassini (1)

1 ISI - Institut des systèmes d'information
2 LISIC - Laboratoire d'Informatique Signal et Image de la Côte d'Opale
3 Computing Science and Mathematics

This article extracts and analyzes local optima networks for the permutation flow-shop problem. Two widely used move operators for permutation representations, namely, swap and insertion, are incorporated into the network landscape model. The performance of a heuristic search algorithm on this problem is also analyzed. In particular, we study the correlation between local optima network features and the performance of an iterated local search heuristic. Our analysis reveals that network features can explain and predict problem difficulty. The evidence confirms the superiority of the insertion operator for this problem.

Domaines: Intelligence artificielle [cs.AI]

Fabio Daolio, Sébastien Verel, Gabriela Ochoa, Marco Tomassini. Local Optima Networks of the Permutation Flow-Shop Problem. International Conference on Artificial Evolution (EA 2013), Oct 2013, Bordeaux, France. ⟨hal-00945475⟩