Beyond permutation equivariance in graph networks

2021/03/25 by Emma Slade, Slade, Emma, Francesco Farina +1 · 1 citation
Biochemistry, Genetics and Molecular Biology · Computer Science · #Advanced Graph Neural Networks #Bioinformatics and Genomic Networks #Epigenetics and DNA Methylation #FOS: Computer and information sciences #Machine Learning (cs.LG)

paper · pdf · doi:10.48550/arxiv.2103.14066

openalex publication_date 2021/03/25 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

In this draft paper, we introduce a novel architecture for graph networks which is equivariant to the Euclidean group in n-dimensions. The model is designed to work with graph networks in their general form and can be shown to include particular variants as special cases. Thanks to its equivariance properties, we expect the proposed model to be more data efficient with respect to classical graph architectures and also intrinsically equipped with a better inductive bias. We defer investigating this matter to future work.

Citations

• Neural Message Passing for Quantum Chemistry
• Tensor field networks: Rotation- and translation-equivariant neural networks for 3D point clouds
• E(n) Equivariant Graph Neural Networks
• Autobahn: Automorphism-based Graph Neural Nets
• A Comprehensive Survey on Graph Neural Networks

Cited by

• Symmetry-driven graph neural networks

Related

• Data efficiency in graph networks through equivariance
• Revisiting Multi-Permutation Equivariance through the Lens of Irreducible Representations
• Invariant and Equivariant Graph Networks
• Subgraph Permutation Equivariant Networks
• Permutation Equivariant Generative Adversarial Networks for Graphs