A Multi-Task Optimization-based Heterogeneous Expensive Multi-Objective Evolutionary Algorithm

Chunlin He,Yong Zhang

DOI: https://doi.org/10.1109/aiotc63215.2024.10748310

2024-01-01

Abstract:Multi-objective optimization problems with different evaluation costs for performance metrics or objective values are commonly seen in the real world. Such problems are known as heterogeneous expensive multi-objective optimization problems (HE-MOPs). Although there are some researches available, the heterogeneity among objectives can cause search bias, which affects the performance of these methods on HEMOPs. To address this, we propose a Multi-Task Optimizationbased Heterogeneous Expensive Multi-Objective Evolutionary Algorithm (MT-HEMOEA). This algorithm firstly uses a multilayer sampling method based on objective evaluation costs to obtain a uniformly distributed sample set on objectives space. Then we employ multi-task exploitation strategy focused on convergence and multi-task exploration strategy focused on diversity to reduce search bias and improve sample diversity. Experimental results on WFG test problems show that the proposed algorithm achieves higher-quality optimal solution sets compared to four existing heterogeneous multi-objective optimization algorithms.