Scholarlink

This preliminary study adapts the CRS4EA framework to evaluate dynamic metaheuristic optimization algorithms (DMAs) in dynamic optimization using a Glicko-2-based chess rating system. The DMAs are modeled as players in round-robin tournaments across multiple environments, with performance assessed via pairwise comparisons of fitness values of their best solutions in each environment. The ratings and rating deviation (RD) are updated iteratively without requiring global optimum knowledge, ensuring robust, statistically sound rankings. The framework’s flexibility supports comparisons across diverse problems and environments, offering a reliable alternative to traditional statistical tests. The preliminary findings indicate that CRS4EA achieves stability and performance comparable to established statistical methods, paving the way for further refinements in DMA evaluation.

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

Chess Rating-Based Evaluation of Dynamic Optimization Metaheuristic Algorithms

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Matej Črepinšek,
  • Matej Moravec,
  • Miha Ravber

摘要

This preliminary study adapts the CRS4EA framework to evaluate dynamic metaheuristic optimization algorithms (DMAs) in dynamic optimization using a Glicko-2-based chess rating system. The DMAs are modeled as players in round-robin tournaments across multiple environments, with performance assessed via pairwise comparisons of fitness values of their best solutions in each environment. The ratings and rating deviation (RD) are updated iteratively without requiring global optimum knowledge, ensuring robust, statistically sound rankings. The framework’s flexibility supports comparisons across diverse problems and environments, offering a reliable alternative to traditional statistical tests. The preliminary findings indicate that CRS4EA achieves stability and performance comparable to established statistical methods, paving the way for further refinements in DMA evaluation.