Scholarlink

This chapter explores Large-Scale Optimization Problems (LSOPs), characterized by high-dimensional search spaces, complex constraints, and nonlinear interactions, making traditional Evolutionary Computation (EC) algorithms inefficient due to the curse of dimensionality. To address these challenges, the chapter introduces Cooperative Coevolution (CC), which decomposes LSOPs into smaller subproblems, and various grouping strategies such as Differential Grouping (DG), Recursive Differential Grouping (RDG), and Extended Recursive Differential Grouping (ERDG) for handling variable interactions. Non-grouping methods like Competitive Swarm Optimization (CSO) and multiple population differential evolution (mDE-brM) are considered alternatives for maintaining search diversity in high-dimensional spaces. Real-world applications, including the Large-Scale Traveling Salesman Problem (LSTSP) and Feature Selection in Machine Learning, demonstrate the practical significance of LSOPs in logistics, AI, and smart systems. By integrating evolutionary strategies, hybrid approaches, and problem decomposition, this chapter provides an effective framework for solving LSOPs, supporting advancements in AI-driven optimization, intelligent decision-making, and green energy systems.

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

EC for Large-Scale Optimization: Methods and Real-World Optimization Problems

  • 章节

  • 发表日期：

    2026年

查看全文
  • Yuefeng Xu,
  • Rui Zhong,
  • Junqi Zhang,
  • Chao Zhang,
  • Jun Yu

摘要

This chapter explores Large-Scale Optimization Problems (LSOPs), characterized by high-dimensional search spaces, complex constraints, and nonlinear interactions, making traditional Evolutionary Computation (EC) algorithms inefficient due to the curse of dimensionality. To address these challenges, the chapter introduces Cooperative Coevolution (CC), which decomposes LSOPs into smaller subproblems, and various grouping strategies such as Differential Grouping (DG), Recursive Differential Grouping (RDG), and Extended Recursive Differential Grouping (ERDG) for handling variable interactions. Non-grouping methods like Competitive Swarm Optimization (CSO) and multiple population differential evolution (mDE-brM) are considered alternatives for maintaining search diversity in high-dimensional spaces. Real-world applications, including the Large-Scale Traveling Salesman Problem (LSTSP) and Feature Selection in Machine Learning, demonstrate the practical significance of LSOPs in logistics, AI, and smart systems. By integrating evolutionary strategies, hybrid approaches, and problem decomposition, this chapter provides an effective framework for solving LSOPs, supporting advancements in AI-driven optimization, intelligent decision-making, and green energy systems.