Parallel Multi-objective Gene Expression Programming Based on Area Penalty

Evolutionary algorithms are particularly desirable to solve multi-objective optimization problems. To improve the evolutionary efficiency, parallel multi-objective gene expression programming based on area penalty (PGEP-AP) is proposed in this paper. The main contributions include: (1) proposing the parallel multi-objective gene expression programming (GEP) to improve the searching efficiency, (2) applying area penalty strategy to avoid the appearance of over-lapping of each evolution subspace and reduce the convergence time of each parallel subpopulation, (3) applying individual migration strategy to improve the total parallel searching speed. Experimental results suggest that PGEP-AP can obtain much more high-quality and evenly distributed nondominated Pareto solutions compared with SPEA, PAES, and NSGA-II.