An ε-Domination based Two-Archive 2 Algorithm for Many-Objective Optimization

The two-archive 2 algorithm (Two_Arch2) is a many-objective evolutionary for balancing the convergence, diversity, and complexity using diversity archive (DA) convergence (CA). However, individuals in DA are selected based on traditional Pareto dominance which decreases selection pressure high-dimensional problems. even cannot converge due to weak pressure. Meanwhile, Two_Arch2 adopts as output of hard maintain coverage final solutions synchronously increase algorithm. To improve distribution solutions, an ε-domination (ε-Two_Arch2) problems (MaOPs) proposed this paper. In ε- Two_Arch2, decrease computational speed up novel framework with fast update strategy proposed; pressure, assigned DA; guarantee uniform solution, boundary protection I ε + indicator designated two steps strategies CA. evaluate performance algorithm, series benchmark functions different numbers objectives solved. results demonstrate that method competitive state-of-the-art multi-objective algorithms efficiency significantly improved compared Two_Arch2.