Scholarlink

<p>To transfer effective knowledge between tasks, improve the effectiveness of intertask interactions and enhance the performance of evolutionary multitasking optimization algorithms, this article presents an adaptive evolutionary multitasking optimization algorithm based on a multitransfer strategy (EMTO-MS). EMTO-MS uses two transfer strategies for intertask knowledge transfer and adaptively allocates computational resources to these transfer strategies based on the performance of each strategy during the evolutionary process. In addition, the strength of interaction between two tasks is adaptively adjusted according to the transfer success rate and the similarity between the two tasks during the optimization process. To evaluate the performance of EMTO-MS, we compared EMTO-MS with nine mainstream evolutionary multitasking algorithms on the single-objective multitasking benchmarks,and with five outstanding EMTO algorithms on a complex multitask benchmark. The experimental results demonstrate that the proposed EMTO-MS outperforms most of the compared algorithms in terms of search accuracy and convergence.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Multitasking optimization algorithm based on a multitransfer strategy

  • 期刊论文

  • 发表日期：

    2025年4月15日

查看全文
  • Xiaoyu Li,
  • Lei Wang,
  • Qiaoyong Jiang

摘要

To transfer effective knowledge between tasks, improve the effectiveness of intertask interactions and enhance the performance of evolutionary multitasking optimization algorithms, this article presents an adaptive evolutionary multitasking optimization algorithm based on a multitransfer strategy (EMTO-MS). EMTO-MS uses two transfer strategies for intertask knowledge transfer and adaptively allocates computational resources to these transfer strategies based on the performance of each strategy during the evolutionary process. In addition, the strength of interaction between two tasks is adaptively adjusted according to the transfer success rate and the similarity between the two tasks during the optimization process. To evaluate the performance of EMTO-MS, we compared EMTO-MS with nine mainstream evolutionary multitasking algorithms on the single-objective multitasking benchmarks,and with five outstanding EMTO algorithms on a complex multitask benchmark. The experimental results demonstrate that the proposed EMTO-MS outperforms most of the compared algorithms in terms of search accuracy and convergence.