Incorporando técnicas de mineração de dados à metaheurística GRASP

Abstract

In this work, we investigate the efficiency of incorporating Data Mining techniques to the GRASP metaheuristic in order to introduce memory to this strategy, generating a hybrid version of GRASP, called GRASP-MD. The Set Packing Problem (SPP) was used to validate this proposal and different versions of the hybrid metaheuristic were tested and analyzed. Computational experiments, comparing traditional GRASP and different hybrid approaches, showed that employing patterns mined from an elite set of solutions conducted to better results. Besides, additional performance experiments evidenced that data mining strategies accelerate the process of finding good solutions.