提出可扩展的高阶图学习框架，提升表达能力同时保持高效计算。
Scaling Higher-Order Graph Learning with Maximal Clique Complexes
- 基于极大团复形构建高效高阶图模型，避免显式枚举团。
- 引入CliqueWalk随机游走方法，计算复杂度线性增长。
- 在保持强性能前提下，显著降低时间和内存开销，适合大规模图分析。
图神经网络（GNN）仅能建模成对交互，而基于细胞复形的高阶模型虽表达能力强，却常面临可扩展性差的问题。本文提出简化的和分解式的细胞魏斯费勒-莱曼测试（sCWL 和 fCWL），在保留CW L测试表达能力的同时提升计算效率。进一步引入极大团复形，使基于细胞复形的神经网络（CWNs）具备更低的时间与内存复杂度，同时保持优异的实证表现。为避免显式枚举团，提出CliqueWalk——一种偏置随机游走算法，可线性缩放于图大小，实现高效最大团采样。这些贡献共同构建了一个可扩展的拓扑学习框架，适用于高阶图表示学习。
原文摘要 · Abstract (English)
Graph neural networks (GNNs) are limited to modeling pairwise interactions, while higher-order models based on cell complexes achieve greater expressivity but often suffer from poor scalability. We introduce simplified and factored cellular Weisfeiler Leman tests (sCWL and fCWL), which preserve the expressivity of the CWL test while improving computational efficiency. We further introduce the maximal clique complex, enabling scalable CWNs with reduced time and memory complexity while retaining strong empirical performance. To avoid explicit clique enumeration, we propose CliqueWalk, a biased random walk that samples maximal cliques and scales linearly with graph size. These contributions yield a scalable topological learning framework for higher-order graph representation.
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。