Breaking Symmetry Bottlenecks in GNN Readouts

2026-02-05

AI Summary

This work addresses a fundamental limitation in the expressive power of graph neural networks (GNNs) that arises not only from message passing but also from the use of linear permutation-invariant readout operations—such as summation or averaging—which discard symmetry-aware information. Drawing on finite-dimensional representation theory, the authors show that such readouts project node embeddings onto the invariant subspace of the permutation group, collapsing nontrivial symmetric structures. To overcome this, they propose a novel readout architecture based on projection decomposition and nonlinear invariant statistics, which preserves symmetry-channel information lost by conventional methods. Replacing only the readout module enables a fixed encoder to distinguish Weisfeiler–Lehman indistinguishable graph pairs and yields significant performance gains across multiple graph learning benchmarks, underscoring the critical role of readout design in GNN expressivity.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Reasoning under Uncertainty: Graphical Models
• Computer Vision: Representation Learning for Vision

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Graph neural networks (GNNs) are widely used for learning on structured data, yet their ability to distinguish non-isomorphic graphs is fundamentally limited. These limitations are usually attributed to message passing; in this work we show that an independent bottleneck arises at the readout stage. Using finite-dimensional representation theory, we prove that all linear permutation-invariant readouts, including sum and mean pooling, factor through the Reynolds (group-averaging) operator and therefore project node embeddings onto the fixed subspace of the permutation action, erasing all non-trivial symmetry-aware components regardless of encoder expressivity. This yields both a new expressivity barrier and an interpretable characterization of what global pooling preserves or destroys. To overcome this collapse, we introduce projector-based invariant readouts that decompose node representations into symmetry-aware channels and summarize them with nonlinear invariant statistics, preserving permutation invariance while retaining information provably invisible to averaging. Empirically, swapping only the readout enables fixed encoders to separate WL-hard graph pairs and improves performance across multiple benchmarks, demonstrating that readout design is a decisive and under-appreciated factor in GNN expressivity.

Problem

• graph neural networks
• readout
• permutation invariance
• symmetry
• expressivity

Innovation

• graph neural networks
• readout
• symmetry
• permutation invariance
• representation theory