Population-Based Incremental Learning with Immigrants Schemes in Changing Environments

The population-based incremental learning (PBIL) algorithm is a combination of evolutionary optimization and competitive learning. PBIL has been successfully applied to dynamic optimization problems (DOPs). It is well known that maintaining the population diversity is important for PBIL to adapt well to dynamic changes. However, PBIL faces a serious challenge when applied to DOPs because at early stages of the optimization process the population diversity is decreased significantly. It has been shown that random immigrants can increase the diversity level maintained by PBIL algorithms and enhance their performance on some DOPs. In this paper, we integrate elitism-based and hybrid immigrants into PBIL to address slightly and severely changing DOPs. Based on a series of dynamic test problems, experiments are conducted to investigate the effect of immigrants schemes on the performance of PBIL. The experimental results show that the integration of elitism-based and hybrid immigrants with PBIL always improves the performance when compared with a standard PBIL on different DOPs. Finally, the proposed PBILs are compared with other peer evolutionary algorithms and show competitive performance.