Section: New Results

Self-adaptive method for a three-objective vector-packing problem

Participants: Nadia Dahmani, François Clautiaux, El-Ghazali Talbi

We introduced a new multi-objective packing problem (MOBPP), in which we optimize the number of bins, the maximum weight of a bin, and the loading balancing. We studied the impact and the combination of two complementary decoding strategies for this problem. A feature of our work is to insert the parameters of the decoders in the representation of the solution. It leads to self-adaptive meta-heuristics, where the algorithm iteratively adapts the parameters during the search. We embedded our approaches in a local search and an evolutionary algorithm for the MOBPP. A comprehensive set of experiments were performed on various benchmarks inspired from the literature. Results confirm that our methods lead to more effective muti-objective metaheuristics for this problem.