A Collaborative Evolutionary Algorithm Based on Decomposition and Dominance for Many-Objective Knapsack Problems

Multi-objective evolutionary algorithms (MOEAs) are popular for solving many-objective knapsack problems. Among various MOEAs, the multi-objective evolutionary algorithm based on decomposition (MOEA/D) behaves well. However, MOEA/D often retains multiple copies of one individual in the population, which might hamper the diversity of the population. To overcome the disadvantage, a collaborative evolutionary algorithm based on decomposition and dominance, called MOEA/D-DDC, is presented in this paper. It mainly adopts a decomposition-dominance collaboration mechanism. The mechanism consists of a decomposition-based population and a dominance-based archive. The decomposition-based population collects elite individuals for the dominance-based archive. Meanwhile the dominance-based archive assists to repair the decomposition-based population and heighten the diversity. The experiment results show that MOEA/D-DDC obtains the better set of solutions than MOEA/D for many-objective knapsack problems with 4 to 8 objectives.