Multiple Tasks for Multiple Objectives: A New Multiobjective Optimization Method via Multitask Optimization

Handling conflicting objectives and finding multiple Pareto optimal solutions are two challenging issues in solving multiobjective optimization problems (MOPs). Inspired by the efficiency of multitask (MTO) problem (MTOP), we propose to treat MOP as a MTOP solve it using MTO. By transforming into MTOP, not only that difficulty handling can be avoided, but also MTO help efficiently find well-distributed for MOP. With above idea, this paper proposes new method via MTO, with following three contributions. Firstly, theorem is proposed theoretically show relationship between how transformed MTOP. Secondly, based on theoretical analysis, tasks (MTMO) framework efficiently. Thirdly, MTMO-based evolutionary algorithm developed MOP, together novel strategies. One target point estimation strategy automatically accurately. The other an archive-based implicit knowledge transfer transferring across enhance results together. superiority validated extensive experiments 15 MOPs objective numbers varying from 3 20 six state-of-the-art algorithms competitors. Therefore, even many-objective new, promising, efficient method.