提出无需依赖同质性的图域自适应方法，提升异质图上的迁移性能。
Graph Domain Adaptation via Homophily-Agnostic Reconstructing Structure
- 分治重构源与目标图的同质/异质版本，分别对齐知识
- 在5个基准数据集上优于现有方法，异质图上提升显著
- 适合标签稀缺且同质性未知的图迁移场景
图域自适应（GDA）将标注的源图知识迁移到无标签的目标图，以缓解标签稀缺问题。然而，现有方法通常假设源图与目标图均具有同质性，导致在异质图上表现不佳。由于目标图无标签，其同质性水平无法预先判断。为此，我们提出一种同质性无关的新方法，有效实现不同同质性程度图之间的知识迁移。具体而言，采用分治策略，分别重建源图与目标图的高同质性和高异质性变体，并在对应变体间进行知识对齐。在五个基准数据集上的大量实验表明，该方法性能优异，尤其在异质图上优势明显。
原文摘要 · Abstract (English)
Graph Domain Adaptation (GDA) transfers knowledge from labeled source graphs to unlabeled target graphs, addressing the challenge of label scarcity. However, existing GDA methods typically assume that both source and target graphs exhibit homophily, leading existing methods to perform poorly when heterophily is present. Furthermore, the lack of labels in the target graph makes it impossible to assess its homophily level beforehand. To address this challenge, we propose a novel homophily-agnostic approach that effectively transfers knowledge between graphs with varying degrees of homophily. Specifically, we adopt a divide-and-conquer strategy that first separately reconstructs highly homophilic and heterophilic variants of both the source and target graphs, and then performs knowledge alignment separately between corresponding graph variants. Extensive experiments conducted on five benchmark datasets demonstrate the superior performance of our approach, particularly highlighting its substantial advantages on heterophilic graphs.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。