Scholarlink

<p>Social Learning Particle Swarm Optimization (SLPSO) is an advanced variant of the PSO algorithm, designed to enhance optimization performance in Large-Scale Global Optimization (LSGO) problems. However, SLPSO encounters significant challenges, particularly in maintaining a balanced trade-off between exploration and exploitation, which limits its effectiveness in complex optimization tasks. In response to these limitations, this paper proposes the Pulse-based Collective Learning Swarm Optimizer (PCLSO). The PCLSO introduces two key innovations: collective learning, which replaces the traditional average position updates in SLPSO, thereby enhancing exploration by leveraging knowledge from multiple high-quality particles; and a pulse-strategy, which addresses convergence stagnation by dynamically perturbing the global best solution to improve exploitation capabilities. Extensive experimental evaluations on the CEC’2010 and CEC’2013 benchmark suites demonstrate that PCLSO significantly outperforms SLPSO, as well as several advanced and classical PSO and DE variants, and the latest cooperative coevolutionary (CC)-based algorithms. Furthermore, to validate its practical applicability, PCLSO is applied to the biological Multiple Sequence Alignment (MSA) problem using Hidden Markov Models (HMM), where it exhibits superior performance compared to other state-of-the-art metaheuristic algorithms. In summary, the study presents a robust optimization tool that advances the field of LSGO and shows promise for addressing complex real-world optimization challenges. The source code of the PCLSO algorithm is publicly available at <a href="https://github.com/tsingke/PCLSO">https://github.com/tsingke/PCLSO</a>.</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

Pulse-strategy collective learning swarm optimizer for large-scale global optimization

  • 期刊论文

  • 发表日期：

    2025年6月19日

查看全文
  • Xiaoyu Liu,
  • Qingke Zhang,
  • Shuzhao Pang,
  • Jiajun Sun,
  • Huaxiang Zhang

摘要

Social Learning Particle Swarm Optimization (SLPSO) is an advanced variant of the PSO algorithm, designed to enhance optimization performance in Large-Scale Global Optimization (LSGO) problems. However, SLPSO encounters significant challenges, particularly in maintaining a balanced trade-off between exploration and exploitation, which limits its effectiveness in complex optimization tasks. In response to these limitations, this paper proposes the Pulse-based Collective Learning Swarm Optimizer (PCLSO). The PCLSO introduces two key innovations: collective learning, which replaces the traditional average position updates in SLPSO, thereby enhancing exploration by leveraging knowledge from multiple high-quality particles; and a pulse-strategy, which addresses convergence stagnation by dynamically perturbing the global best solution to improve exploitation capabilities. Extensive experimental evaluations on the CEC’2010 and CEC’2013 benchmark suites demonstrate that PCLSO significantly outperforms SLPSO, as well as several advanced and classical PSO and DE variants, and the latest cooperative coevolutionary (CC)-based algorithms. Furthermore, to validate its practical applicability, PCLSO is applied to the biological Multiple Sequence Alignment (MSA) problem using Hidden Markov Models (HMM), where it exhibits superior performance compared to other state-of-the-art metaheuristic algorithms. In summary, the study presents a robust optimization tool that advances the field of LSGO and shows promise for addressing complex real-world optimization challenges. The source code of the PCLSO algorithm is publicly available at https://github.com/tsingke/PCLSO.