对比8种位置编码在图模型中的表现，发现越复杂的编码不等于越好。
Benchmarking Positional Encodings for GNNs and Graph Transformers
- 构建统一框架，分离编码与模型结构，公平比较9种位置编码
- 实测显示高表达力编码常无效甚至拖累性能，尤其在真实数据上
- 发现若干简单组合能媲美甚至超越最新顶尖方法，适合实用研究
位置编码（PEs）对图神经网络（GNNs）和图Transformer注入结构信息至关重要，但其实际影响仍不明确。本文提出一个统一的基准测试框架，将位置编码与模型架构解耦，实现对8种GNN与Transformer模型、9种位置编码及10个合成与真实数据集的公平比较。在超过500种模型-编码-数据组合中，发现常用的表达能力代理（如Weisfeiler-Lehman可区分性）无法可靠预测下游性能。特别是，高表达力的位置编码在真实任务中常无法提升性能，甚至导致下降。同时，我们识别出若干简单且被忽视的模型-编码组合，其表现可匹配或超越近期最先进方法。结果表明位置编码效果高度依赖任务，强调需通过实证而非理论表达力来评估。为支持可复现研究，本文开源了该基准框架，用于图学习任务中的位置编码评估。
原文摘要 · Abstract (English)
Positional Encodings (PEs) are essential for injecting structural information into Graph Neural Networks (GNNs), particularly Graph Transformers, yet their empirical impact remains insufficiently understood. We introduce a unified benchmarking framework that decouples PEs from architectural choices, enabling a fair comparison across 8 GNN and Transformer models, 9 PEs, and 10 synthetic and real-world datasets. Across more than 500 model-PE-dataset configurations, we find that commonly used expressiveness proxies, including Weisfeiler-Lehman distinguishability, do not reliably predict downstream performance. In particular, highly expressive PEs frequently fail to improve, and can even degrade performance on real-world tasks. At the same time, we identify several simple and previously overlooked model-PE combinations that match or outperform recent state-of-the-art methods. Our results demonstrate the strong task-dependence of PEs and underscore the need for empirical validation beyond theoretical expressiveness. To support reproducible research, we release an open-source benchmarking framework for evaluating PEs for graph learning tasks.
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。