Section: New Results

On the Neutrality of Combinatorial Optimization Problem to Design an Efficient Neutrality-based Local Search

Participants : M.-E. Marmion, C. Dhaenens, L. Jourdan, A. Liefooghe, S. Verel.

In the context of the permutation flowshop scheduling problem, a deep landscape analysis focused on the neutrality property has been driven [hal-00550356]. This analysis characterizes the neutral networks of the local optima in order to make propositions about the way to exploit it in algorithms. Hence, NILS has been designed in order to exploit the neutrality of local optima [hal-00563459]. As soon as a local optimum is found, the search is allowed to move to equivalent neighbors. Moreover, NILS gets only one parameter that controls the number of solutions allowed to be visited with the same fitness value. NILS has been tested on flowshop and has shown promising results.