Large-Scale Evolutionary Multiobjective Optimization Assisted by Directed Sampling

It is particularly challenging for evolutionary algorithms to quickly converge the Pareto front in large-scale multiobjective optimization. To tackle this problem, article proposes a algorithm assisted by some selected individuals generated directed sampling (DS). At each generation, set of closer ideal point chosen performing DS decision space, and those nondominated ones sampled solutions are used assist reproduction improve convergence In addition, elitist sorting adopted complementarily environmental selection with reference vector-based method order maintain diversity population. Our experimental results show that proposed highly competitive on optimization test problems up 5000 variables compared five state-of-the-art algorithms.