Usually, the choice of the best heuristic for a problem is implicitly done by the metaheuristic practitioner who trusts his own expertise to measure the efficiency of different approaches -e.g. to better solve a problem that is in general of large size, he might prefer a specific local search rather than another without a tangible reason. However, these decisions can be partially automated and it is possible to switch between several heuristics during the optimization process to permit to the search process a wide range exploration of possibilities and thus to provide a more robust response to the problem. Hyper-heuristics are thus this class of heuristics which can be seen as heuristics to choose heuristics.

This session is dedicated to modern hyper-heuristics able to efficiently tackle very large size and hard combinatorial problem.