研究人员推出了一种新颖的框架MeEvo，它结合了自然进化和元认知进化，以增强使用大型语言模型（LLMs）的自动启发式设计（AHD）。MeEvo通过自然进化实现启发式代码探索并保留推理痕迹，同时利用元认知进化通过反思来优化这些启发式方法，从而解决了现有方法的局限性。这种双层方法使以种群为驱动的探索和以反思为驱动的优化能够协同工作。实验表明，MeEvo在复杂优化问题上优于当前基于LLM的AHD架构。 AI
影响 MeEvo的组合进化方法有望实现更高效、更稳定的AI驱动的问题解决方法，尤其是在复杂的优化任务中。
排序理由 该集群包含一篇详细介绍自动启发式设计新方法的学术论文。
在 arXiv cs.NE (Neural & Evolutionary) 阅读 →
AI 生成摘要 · Google Gemini · 来自 4 个来源。 我们如何撰写摘要 →
完整方法见我们的编辑标准。
arXiv:2606.14202v1 Announce Type: cross Abstract: Large Language Models (LLMs) have advanced Automatic Heuristic Design (AHD) by enabling heuristic generation through reasoning and code synthesis. Existing LLM-based AHD architectures mainly follow two paradigms: Natural Evolution…
Large Language Models (LLMs) have advanced Automatic Heuristic Design (AHD) by enabling heuristic generation through reasoning and code synthesis. Existing LLM-based AHD architectures mainly follow two paradigms: Natural Evolution, which uses crossover and mutation to explore heu…
Large Language Models (LLMs) have advanced Automatic Heuristic Design (AHD) by enabling heuristic generation through reasoning and code synthesis. Existing LLM-based AHD architectures mainly follow two paradigms: Natural Evolution, which uses crossover and mutation to explore heu…
Large Language Models (LLMs) have advanced Automatic Heuristic Design (AHD) by enabling heuristic generation through reasoning and code synthesis. Existing LLM-based AHD architectures mainly follow two paradigms: Natural Evolution, which uses crossover and mutation to explore heu…