Large-Scale Multi-objective Evolutionary Algorithms Based on Adaptive Immune-Inspirated.

This paper proposes an adaptive immune-inspired algorithm to tackle the issues of insufficient diversity and local optima in large-scale multi-objective optimization problems. The algorithm utilizes immune multi-objective evolutionary algorithm as a framework and adaptively selects two different antibody generation strategies based on the concentration of high-quality antibodies. Among them, one approach utilizes the proportional cloning operator to generate offspring, which ensures convergence speed and population diversity, preventing the algorithm from getting trapped in local optimization. The other approach introduces a competitive learning strategy to guide individuals towards the correct direction in the population. Additionally, the proposed algorithm employs a displacement density-based strategy to determine the antibody status. Experimental results demonstrate that the proposed algorithm outperforms five state-of-the-art multi-objective evolutionary algorithms in large-scale multi-objective optimization problems with up to 500 decision variables.