Parameter analysis on multi‐factorial evolutionary algorithm
Qingzheng Xu, Jianhang Zhang, Rong Fei, Wei Li · The Journal of Engineering · 2020
As a novel and representative multi‐task optimisation (MTO) paradigm, multi‐factorial evolutionary algorithm (MFEA) can solve multiple self‐contained tasks simultaneously. Its overall performance highly depends on control parameters. The aim of this research work is to analyse three parameters, namely, probability of individual learning, probability of intra‐crossover and probability of inter‐crossover, controlled by the user. Experimental results on MTO problems demonstrate the superiority of MFEA with a smaller probability of individual learning in a fair competitive environment. While the influence of probabilities of intra‐crossover and inter‐crossover is unpredictable based on the task's features, the basic selection principle and the optimal value are provided based on massive simulated data.