研究人员开发了一种通过分析其程序化表示的复杂度来表征优化问题的新方法。该方法使用 Halstead 体积和代码熵等度量，这些度量可以快速计算且对变换不变。该研究将这些度量应用于 BBOB 套件和前馈神经网络训练，发现与算法性能呈负相关，表明它们作为算法选择和分析的预测元特征的效用。 AI
影响 这项研究可能导致机器学习任务中更有效的算法选择和配置。
排序理由 该集群包含一篇详细介绍新研究方法的学术论文。
AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →
arXiv:2608.08898v1 Announce Type: cross Abstract: Characterising optimisation problem instances is a fundamental part of understanding the behaviour and performance of different algorithms as well as providing information for algorithm selection and configuration. In this paper w…
Characterising optimisation problem instances is a fundamental part of understanding the behaviour and performance of different algorithms as well as providing information for algorithm selection and configuration. In this paper we propose a novel approach to problem characterisa…