Learning Graph Foundation Models on Riemannian Graph-of-Graphs
2026-05-11 • Machine Learning
Machine Learning
AI summaryⓘ
The authors introduce R-GFM, a new type of graph foundation model that improves how computers understand graphs by looking at their structure at multiple scales instead of just one fixed size. Unlike previous models that only consider a limited neighborhood around each node, R-GFM examines subgraphs at different hop distances and uses math from curved spaces (Riemannian manifolds) to better learn graph features. Their approach helps reduce errors when applying the model to different types of graph data and performs better in experiments on various tasks. This work offers a way to make graph learning more flexible and accurate.
Graph foundation modelsSubgraph samplingRiemannian manifoldsMulti-scale graphsGraph-of-GraphsStructural domain generalizationHop distanceGraph machine learning
Authors
Haokun Liu, Zezhong Ding, Xike Xie
Abstract
Graph foundation models (GFMs), pretrained on massive graph data, have transformed graph machine learning by supporting general-purpose reasoning across diverse graph tasks and domains. Existing GFMs pretrained with fixed-hop subgraph sampling impose a fixed receptive field, causing scale mismatch on diverse tasks, which often require heterogeneous and unknown structural contexts beyond a fixed sampling scale. We propose R-GFM, a Riemannian Graph-of-Graphs (GoG) based foundation model, that treats structural scale as a first-class citizen in modeling. R-GFM constructs a multi-scale GoG over-sampled subgraphs at different hop distances and learns geometry-adaptive representations from Riemannian manifolds. Theoretical analysis shows that R-GFM reduces structural domain generalization error compared to fixed-scale GFMs. Experiments on various datasets demonstrate that R-GFM achieves state-of-the-art performance, with up to a 49% relative improvement on downstream tasks. Our code is available at https://github.com/USTC-DataDarknessLab/R-GFM.