English

DRew: Dynamically Rewired Message Passing with Delay

Machine Learning 2023-05-19 v2 Artificial Intelligence Machine Learning

Abstract

Message passing neural networks (MPNNs) have been shown to suffer from the phenomenon of over-squashing that causes poor performance for tasks relying on long-range interactions. This can be largely attributed to message passing only occurring locally, over a node's immediate neighbours. Rewiring approaches attempting to make graphs 'more connected', and supposedly better suited to long-range tasks, often lose the inductive bias provided by distance on the graph since they make distant nodes communicate instantly at every layer. In this paper we propose a framework, applicable to any MPNN architecture, that performs a layer-dependent rewiring to ensure gradual densification of the graph. We also propose a delay mechanism that permits skip connections between nodes depending on the layer and their mutual distance. We validate our approach on several long-range tasks and show that it outperforms graph Transformers and multi-hop MPNNs.

Keywords

Cite

@article{arxiv.2305.08018,
  title  = {DRew: Dynamically Rewired Message Passing with Delay},
  author = {Benjamin Gutteridge and Xiaowen Dong and Michael Bronstein and Francesco Di Giovanni},
  journal= {arXiv preprint arXiv:2305.08018},
  year   = {2023}
}

Comments

Accepted at ICML 2023; 16 pages

R2 v1 2026-06-28T10:33:49.544Z