Search papers, labs, and topics across Lattice.
The paper addresses the challenge of varying input feature spaces in graph learning, which hinders the development of graph foundation models. They introduce ALL-IN, a method that projects node features into a shared random space and constructs representations using covariance-based statistics, achieving invariance to input feature permutations and orthogonal transformations. Empirical results demonstrate that ALL-IN enables strong generalization performance across diverse graph tasks on unseen datasets with novel input features, without requiring architecture changes or retraining.
Graph models can now generalize to entirely new datasets with different input features, thanks to a simple projection into a shared random space.
Unlike vision and language domains, graph learning lacks a shared input space, as input features differ across graph datasets not only in semantics, but also in value ranges and dimensionality. This misalignment prevents graph models from generalizing across datasets, limiting their use as foundation models. In this work, we propose ALL-IN, a simple and theoretically grounded method that enables transferability across datasets with different input features. Our approach projects node features into a shared random space and constructs representations via covariance-based statistics, thus eliminating dependence on the original feature space. We show that the computed node-covariance operators and the resulting node representations are invariant in distribution to permutations of the input features. We further demonstrate that the expected operator exhibits invariance to general orthogonal transformations of the input features. Empirically, ALL-IN achieves strong performance across diverse node- and graph-level tasks on unseen datasets with new input features, without requiring architecture changes or retraining. These results point to a promising direction for input-agnostic, transferable graph models.
