Multiobjective multitasking optimization assisted by multidirectional prediction method

Abstract Multiobjective multitasking optimization (MTO) is an emerging research topic in the field of evolutionary computation, which has attracted extensive attention, and many (EMT) algorithms have been proposed. One core issues, designing efficient transfer strategy, scarcely explored. Keeping this mind, paper first attempt to design strategy based on multidirectional prediction method. Specifically, population divided into multiple classes by binary clustering method, representative point each class calculated. Then, effective direction method developed generate directions points. Afterward, a mutation strength adaptation proposed according improvement degree class. Finally, predictive transferred solutions are generated as knowledge strengths. By above process, multiobjective EMT algorithm presented. Experiments two MTO test suits indicate that competitive other state-of-the-art algorithms.