English(EN) Not All Nodes Are Created Equal: Homophily-Aware Stratification for Stable GNN Evaluation

新的 HP 方法增强了图神经网络评估的稳定性

作者 PulseAugur 编辑部 · [1 个来源] · 2026-09-18 04:00

研究人员开发了一种名为 HP（同质性感知分层）的新方法，以提高图神经网络 (GNN) 评估的可靠性。传统的 GNN 训练和测试数据随机划分可能导致报告的准确性因划分中邻域同质性的差异而产生显著变化。HP 通过基于节点同质性进行数据分层来解决这个问题，确保不同的测试折叠具有相似的局部关系一致性，并保持类别分布。该方法在 15 个数据集和 7 种 GNN 架构的基准套件中展示了更高的稳定性和可靠性。 AI

影响通过提供更稳定的评估框架，增强了 GNN 研究的可靠性。

排序理由介绍图神经网络新评估方法的学术论文。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv stat.ML 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。我们如何撰写摘要 →

本文如何被排名

Signal score: 21 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 介绍图神经网络新评估方法的学术论文。[lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration: Single-source cluster; Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics: paper, other; Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance: High; Clearly on-topic for AI-industry coverage.
Story freshness: Breaking (< 6h); Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

完整方法见我们的编辑标准。

报道来源 [1]

arXiv stat.ML TIER_1 English(EN) · Naga Venkata Sai Jitin Jami, Thomas Altstidl, Sebastian Hoefler, Jonas Mueller, Dario Zanca, Bjoern Eskofier, Heike Leutheuser · 2026-09-18 04:00

并非所有节点都生而平等：同质性感知分层用于稳定的GNN评估

arXiv:2609.19210v1 Announce Type: cross Abstract: Graph neural networks are widely used for transductive node classification, with accuracy typically measured on randomly drawn train/validation/test splits. Reported accuracy has been shown to shift substantially across different …

本文如何被排名

报道来源 [1]

并非所有节点都生而平等：同质性感知分层用于稳定的GNN评估

相关实体

相关话题