English

Ego-GNNs: Exploiting Ego Structures in Graph Neural Networks

Machine Learning 2021-07-26 v1

Abstract

Graph neural networks (GNNs) have achieved remarkable success as a framework for deep learning on graph-structured data. However, GNNs are fundamentally limited by their tree-structured inductive bias: the WL-subtree kernel formulation bounds the representational capacity of GNNs, and polynomial-time GNNs are provably incapable of recognizing triangles in a graph. In this work, we propose to augment the GNN message-passing operations with information defined on ego graphs (i.e., the induced subgraph surrounding each node). We term these approaches Ego-GNNs and show that Ego-GNNs are provably more powerful than standard message-passing GNNs. In particular, we show that Ego-GNNs are capable of recognizing closed triangles, which is essential given the prominence of transitivity in real-world graphs. We also motivate our approach from the perspective of graph signal processing as a form of multiplex graph convolution. Experimental results on node classification using synthetic and real data highlight the achievable performance gains using this approach.

Keywords

Cite

@article{arxiv.2107.10957,
  title  = {Ego-GNNs: Exploiting Ego Structures in Graph Neural Networks},
  author = {Dylan Sandfelder and Priyesh Vijayan and William L. Hamilton},
  journal= {arXiv preprint arXiv:2107.10957},
  year   = {2021}
}

Comments

Submitted to a special session of IEEE-ICASSP 2021

R2 v1 2026-06-24T04:26:52.193Z