Fully Inductive Node Representation Learning via Graph View Transformation

TL;DR

This paper introduces the view space and Graph View Transformation (GVT) to enable fully inductive node representation learning across diverse graph datasets, achieving superior performance without retraining.

Contribution

The paper proposes the view space and GVT as novel methods for fully inductive graph learning, allowing models to generalize to unseen datasets effectively.

Findings

Recurrent GVT outperforms prior models by +8.93% on OGBN-Arxiv.

Recurrent GVT surpasses 12 GNNs by at least +3.30%.

The view space provides a unified representation for diverse graph data.

Abstract

Generalizing a pretrained model to unseen datasets without retraining is an essential step toward a foundation model. However, achieving such cross-dataset, fully inductive inference is difficult in graph-structured data where feature spaces vary widely in both dimensionality and semantics. Any transformation in the feature space can easily violate the inductive applicability to unseen datasets, strictly limiting the design space of a graph model. In this work, we introduce the view space, a novel representational axis in which arbitrary graphs can be naturally encoded in a unified manner. We then propose Graph View Transformation (GVT), a node- and feature-permutation-equivariant mapping in the view space. GVT serves as the building block for Recurrent GVT, a fully inductive model for node representation learning. Pretrained on OGBN-Arxiv and evaluated on 27 node-classification benchmarks, Recurrent GVT outperforms GraphAny, the prior fully inductive graph model, by +8.93% and surpasses 12 individually tuned GNNs by at least +3.30%. These results establish the view space as a principled and effective ground for fully inductive node representation learning.