Scholarlink

<p>The Sine-Cosine Algorithm (SCA) is a promising metaheuristic optimizer, yet its performance may degrade due to the simple search pattern. To address this limitation, this paper proposes a novel Tri-Subpopulation Sigmoid-Enhanced Sine-Cosine Algorithm (TS-SESCA). TS-SESCA divides the population into three distinct subgroups and adopts different search strategies: a mixed strategy group using both sine and cosine updates, an exploration-focused sine group with amplified steps, and an exploitation-driven cosine group with refined movements. Additionally, a sigmoid-based nonlinear control mechanism is introduced to ensure a better balance between exploration and exploitation. Experimental results on the CEC2020 and CEC2022 benchmark functions demonstrate the superior optimization capability of TS-SESCA compared to existing algorithms. Moreover, TS-SESCA is successfully applied to Gene function prediction tasks, where it optimizes ensemble weights of deep learning models and achieves state-of-the-art predictive performance. These results validate the robustness, adaptability, and practical effectiveness of TS-SESCA in both benchmark optimization and real-world bioinformatics applications.</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

Tri-subpopulation sigmoid-enhanced sine-cosine algorithm and its application to gene function prediction problem

  • 期刊论文

  • 发表日期：

    2025年11月19日

查看全文
  • Yang Cao,
  • Yuefeng Xu,
  • Xingbang Du,
  • Rui Zhong,
  • Jun Yu,
  • Masaharu Munetomo

摘要

The Sine-Cosine Algorithm (SCA) is a promising metaheuristic optimizer, yet its performance may degrade due to the simple search pattern. To address this limitation, this paper proposes a novel Tri-Subpopulation Sigmoid-Enhanced Sine-Cosine Algorithm (TS-SESCA). TS-SESCA divides the population into three distinct subgroups and adopts different search strategies: a mixed strategy group using both sine and cosine updates, an exploration-focused sine group with amplified steps, and an exploitation-driven cosine group with refined movements. Additionally, a sigmoid-based nonlinear control mechanism is introduced to ensure a better balance between exploration and exploitation. Experimental results on the CEC2020 and CEC2022 benchmark functions demonstrate the superior optimization capability of TS-SESCA compared to existing algorithms. Moreover, TS-SESCA is successfully applied to Gene function prediction tasks, where it optimizes ensemble weights of deep learning models and achieves state-of-the-art predictive performance. These results validate the robustness, adaptability, and practical effectiveness of TS-SESCA in both benchmark optimization and real-world bioinformatics applications.