.png){kind=icon}
.png){kind=icon}
Return
Parameter adaptation in multifactorial evolutionary algorithm for many-task optimization
.png){kind=icon}
PRE
AI
DOI：10.1007/s12293-021-00347-4.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
The advent of multifactorial optimization (MFO) has made a wind of change in intelligence computation in general and specifically in evolutionary computing. Based on the implicit parallelism of population-based search, MFO optimizes different problems simultaneously and entirely. However, the randomness of knowledge transfers raises the question of how to diminish harmful interactions among tasks for more effective transfers. In recent years, many proposals have been devised to handle this paradigm and improve existing algorithms. Notwithstanding the diversity in their concept, there are few efforts to solve many-task optimization (MaTO) that contains beyond three tasks. In light of this reason, this paper proposes two algorithms named SA-MFEA and LSA-MFEA for MaTO. Instead of utilizing fixed parameters, SA-MFEA and LSA-MFEA adapt the probability of random mating parameter to reduce negative transfers based on the historical memory of successful rmp. Besides, LSA-MFEA is capable of enhancing the exploitation by linear population size reduction. To examine the efficiency of the two proposed algorithms, experiments on various many-task benchmark problems and comparison with several state-of-the-art algorithms have been conducted. The results demonstrated that SA-MFEA and LSA-MFEA are competitive in terms of quality of solutions, convergence trend, and computation time.
Keywords:
Evolutionary multi-task optimization
Multifactorial evolutionary algorithm
Parameters adaptation strategy
Many-task optimization
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
IF:
2.3
Papers:
453
Citations:
718