.png){kind=icon}
.png){kind=icon}
Return
A multi-objective multi-task evolutionary algorithm based on source task transfer
.png){kind=icon}
PRE
AI
DOI：10.1016/j.asoc.2025.112732.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
In the real world, many optimization problems often do not exist in isolation, they usually have complex interactions and dependencies, and have multiple optimization goals. In order to improve the performance of individual task solving, an evolutionary multi-task multi-objective optimization algorithm (MTMOO) is proposed. However, most of the current evolutionary algorithms are based on the assumption that the prior knowledge (experience in solving optimization problems) is zero, which makes the ability of the algorithm to solve problems cannot be improved with the increase of historical experience, and greatly limits the adaptability and learning ability of the algorithm. In order to overcome this limitation, this paper proposed a multi-objective and multi-task adaptive migration Evolutionary algorithm (MOMFEA-STT). The algorithm constructs the parameter sharing model of historical task and target task online. By identifying the degree of association between different tasks, the intensity of cross-task knowledge transfer is automatically adjusted to maximize the capture, sharing and utilization of common useful knowledge to solve related tasks. In addition, in order to strengthen the exploration and exploitation ability of the algorithm and avoid the problem that the algorithm is easy to fall into the local optimum, the MOMFEA-STT adopts a new method of generation of children, generates children by using spiral search mode, and constantly adjusts the search direction of the algorithm. Experimental results show that the proposed method outperforms the existing algorithms on the multi-task optimization benchmark problems.
Keywords:
Multi-objective optimization
Multi-task optimization
Evolutionary algorithm
Adaptive transfer
Journal
IF:
6.6
Papers:
1.4W
Citations:
4.8W
Organization
No organization information available