New Initialisation Techniques for Multi-objective Local Search
Résumé
Given the availability of high-performing local search (LS) for single-objective (SO) optimisation problems, one successful approach to tackle their multi-objective (MO) counterparts is scalarisation-based local search (SBLS). SBLS strategies solve multiple scalarisations, i.e., aggregations of the multiple objectives into a single scalar value, with varying weights. They have been shown to work specially well as the initialisation phase of other types of multi-objective local search, such as Pareto local search (PLS). A major drawback of existing SBLS strategies is that the underlying SO optimiser is unaware of the MO nature of the problem and only returns a single solution, discarding any intermediate solutions that may be of interest. We propose here two new SBLS strategies (ChangeRestart and Change-Direction) that overcome this drawback by augmenting the underlying SO-LS method with an archive of nondominated solutions that is used to dynamically update the scalarisations. The new strategies produce better results on the bi-objective permutation flowshop problem than other five SBLS strategies from the literature, not only on their own but also when used as the initialisation phase of PLS.
Origine | Fichiers produits par l'(les) auteur(s) |
---|
Loading...