Scholarlink

<p>Nowadays, numerous sciences, engineering and technology problems exhibit the nature of large-scale many-objective optimization problems (LSMaOPs). Addressing such LSMaOPs, traditional optimization algorithm faces performance challenges. To overcome the performance challenges, several large-scale many-objective optimization algorithms (LSMaOAs) have been created. Even though the recently developed some LSMaOAs have demonstrated promising performance, there remains a need for further enhancement and improvement in their frameworks to ensue better applicability across more complex and diverse LSMaOPs. In this study, we enhanced the framework of the Competitive Swarm Optimizer (CSO) by incorporating Information Feedback Model (IFM) and Reference Point Based Selection (RPS) strategy, and proposed a novel CSO-IFM-RPS algorithm for LSMaOPs. The proposed CSO-IFM-RPS is assessed on benchmark suites (i.e., LSMOP1-9) with up to 6 objectives and 1000 decision variables, and results also compared with existing approaches, including LMEA, IFM-NSGA-III, CCSO, and S3-CMA-ES. Experimental results reveal that CSO-IFM-RPS achieves statistically significant improvements over existing algorithms, in terms of IGD and Hypervolume. These results confirm the effectiveness of IFM and RPS strategies to enhance performance of CSO in solving complex, LSMaOPs. Furthermore, a real-world validation on the Multi-Objective Traveling Salesman Problem (MOTSP) confirms the generality and utility of the proposed framework.</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

Improving competitive swarm optimizer with information feedback model and reference point based selection for complex optimization problems

  • 期刊论文

  • 发表日期：

    2026年3月17日

查看全文
  • Ritika Chaudhary,
  • Amarjeet Prajapati

摘要

Nowadays, numerous sciences, engineering and technology problems exhibit the nature of large-scale many-objective optimization problems (LSMaOPs). Addressing such LSMaOPs, traditional optimization algorithm faces performance challenges. To overcome the performance challenges, several large-scale many-objective optimization algorithms (LSMaOAs) have been created. Even though the recently developed some LSMaOAs have demonstrated promising performance, there remains a need for further enhancement and improvement in their frameworks to ensue better applicability across more complex and diverse LSMaOPs. In this study, we enhanced the framework of the Competitive Swarm Optimizer (CSO) by incorporating Information Feedback Model (IFM) and Reference Point Based Selection (RPS) strategy, and proposed a novel CSO-IFM-RPS algorithm for LSMaOPs. The proposed CSO-IFM-RPS is assessed on benchmark suites (i.e., LSMOP1-9) with up to 6 objectives and 1000 decision variables, and results also compared with existing approaches, including LMEA, IFM-NSGA-III, CCSO, and S3-CMA-ES. Experimental results reveal that CSO-IFM-RPS achieves statistically significant improvements over existing algorithms, in terms of IGD and Hypervolume. These results confirm the effectiveness of IFM and RPS strategies to enhance performance of CSO in solving complex, LSMaOPs. Furthermore, a real-world validation on the Multi-Objective Traveling Salesman Problem (MOTSP) confirms the generality and utility of the proposed framework.