The AIS-HSL optimizer: An artificial immune system with heuristic social learning

This paper proposes an artificial immune system with heuristic social learning (AIS-HSL) for optimization. In the AIS-HSL optimization, the candidate antibodies is separated into two swarms i.e., the elitist swarm (ES) and the common swarm (CS). Different swarms experience different mutation processes, i.e., a self-learning strategy is required for ES, while a heuristic social-learning (HSL) mechanism is applied to CS. In the HSL mechanism, each antibody in CS learns from a selected antibody in ES based on the probability determined by its affinity to avoid falling into the local optima. Some comparative numerical simulations are arranged to evaluate the performance of the proposed AIS-HSL. The results demonstrate that the proposed AIS-HSL outperforms the canonical opt-aiNet optimization, the IA-AIS optimization and the AAIS-2S optimization in convergence speed and solution accuracy.