AdaSpec: Adaptive Spectrum for Enhanced Node Distinguishability

ICLR 2026 Conference Submission15636 Authors

Keywords: spectral graph neural networks, symmetry on graph, permutation equivalence, node distinguishability

TL;DR: An adaptive plug-in graph matrix generation module to enhance the node distinguishability of spectral graph neural networks.

Abstract: Spectral Graph Neural Networks (GNNs) achieve strong performance in node classification, yet their node distinguishability remains poorly understood. We analyze how graph matrices and node features jointly influence node distinguishability. Further, we derive a theoretical lower bound on the number of distinguishable nodes, which is governed by two key factors: distinct eigenvalues in the graph matrix and nonzero frequency components of node features in the eigenbasis. Based on these insights, we propose AdaSpec, an adaptive graph matrix generation module that enhances node distinguishability of spectral GNNs without increasing the order of computational complexity. We prove that AdaSpec preserves permutation equivariance, ensuring that reordering the graph nodes results in a corresponding reordering of the node embeddings. Experiments across eighteen benchmark datasets validate AdaSpec's effectiveness in improving node distinguishability of spectral GNNs.

Primary Area: learning on graphs and other geometries & topologies

Submission Number: 15636