研究人员开发了一种名为标签上下文分类器（LCC）的新型分类器，以改进异质图中的节点分类。当前的图神经网络（GNN）在处理这类节点具有不同标签连接更紧密的异质图时面临挑战。LCC通过四种类型的游走生成标签上下文嵌入，捕捉高阶类标签连通性，从而解决这一问题。当与现有的GNN集成时，LCC在异质有向图上的性能优于最先进的方法。 AI
影响 提高了异质图中节点分类的准确性，可能改进推荐系统和社交网络分析等领域的应用。
排序理由 该集群包含一篇详细介绍图神经网络新方法的学术论文。
AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →
arXiv:2606.07475v1 Announce Type: cross Abstract: Node classification in graph neural networks (GNNs) has been widely applied in various fields of graph analysis. GNNs achieve high-accuracy node classification in homophilous graphs, where nodes with the same class label tend to b…
Node classification in graph neural networks (GNNs) has been widely applied in various fields of graph analysis. GNNs achieve high-accuracy node classification in homophilous graphs, where nodes with the same class label tend to be connected. However, their performance remains li…