A surrogate assisted evolutionary multitasking optimization algorithm

Evolutionary algorithms (EAs) have been applied with strong abilities to solve a wide range of applications, but it can solve one problem at a time. To improve efficiency, an emerging research paradigm in the field of evolutionary computation, Evolutionary multi-tasking (EMT) was proposed. EMT solves multiple optimization tasks simultaneously. The effectiveness of EMT is to improve the solutions for each task via inter-task knowledge transfer. Multifactorial evolutionary algorithms (MFEAs) is the first algorithm proposed to solve multi-task optimization problems. However, it tends to suffer from the issue of negative knowledge transfer. To address this issue and improve the performance of MFEA, we propose to construct a surrogate model as a helper task is optimized and target task simultaneously in MFEA. According to the proposed method, the surrogate model is a related task for each corresponding target task to enhance positive inter-task knowledge transfer. Besides, the surrogate model can reduce the number of local optima and has a simple structure. Experiments are conducted on benchmarks and real-world reservoir flood generation power problems to examine the performance of the proposed algorithm. Comparative experiments on several widely used test problems demonstrated that surrogate models as helper tasks enable significantly improve the performance of MFEA.