A Decomposition based estimation of distribution algorithm for multiobjective knapsack problems
Yang Li, Aimin Zhou, Guixu Zhang · 2012
Multiobjective knapsack problems (MOKPs) are useful for both theoretical studies and practical applications. This paper proposes a novel algorithm, named multiobjective estimation of distribution algorithm based on decomposition (MEDA/D), for dealing with MOKPs. In MEDA/D, a probabilistic model based offspring reproduction operator is incorporated into the multiobjective evolutionary algorithm based on decomposition (MOEA/D). The population is maintained by the MOEA/D framework and new solutions are sampled from the probabilistic models. MEDA/D is applied to a set of test instances and compared with an MOEA/D with generic crossover/mutation operators. The statistical results show that the new approach is promising for dealing with MOKPs.