提出新型消息传递机制，缓解图神经网络过平滑问题。
Heterophily-informed Message Passing
- 通过学习嵌入动态调节消息聚合方式
- 在多个数据集上提升分类与分子生成性能
- 无需额外标签，适合生成建模等场景
图神经网络因隐含同质性假设，易受过平滑影响。本文提出一种新方案，通过局部调节消息聚合类型与范围，保留信息的低频与高频成分。方法仅依赖学习到的嵌入，无需辅助标签，使异质性感知嵌入的优势扩展至生成建模等更广泛场景。实验在多种数据集和GNN架构上验证了性能提升，并揭示了标准分类基准中的异质性模式。应用于分子生成时，在化学生物学基准上取得显著改进。
原文摘要 · Abstract (English)
Graph neural networks (GNNs) are known to be vulnerable to oversmoothing due to their implicit homophily assumption. We mitigate this problem with a novel scheme that regulates the aggregation of messages, modulating the type and extent of message passing locally thereby preserving both the low and high-frequency components of information. Our approach relies solely on learnt embeddings, obviating the need for auxiliary labels, thus extending the benefits of heterophily-aware embeddings to broader applications, e.g., generative modelling. Our experiments, conducted across various data sets and GNN architectures, demonstrate performance enhancements and reveal heterophily patterns across standard classification benchmarks. Furthermore, application to molecular generation showcases notable performance improvements on chemoinformatics benchmarks.
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。