English

Self-Attention in Colors: Another Take on Encoding Graph Structure in Transformers

Machine Learning 2023-04-24 v1

Abstract

We introduce a novel self-attention mechanism, which we call CSA (Chromatic Self-Attention), which extends the notion of attention scores to attention _filters_, independently modulating the feature channels. We showcase CSA in a fully-attentional graph Transformer CGT (Chromatic Graph Transformer) which integrates both graph structural information and edge features, completely bypassing the need for local message-passing components. Our method flexibly encodes graph structure through node-node interactions, by enriching the original edge features with a relative positional encoding scheme. We propose a new scheme based on random walks that encodes both structural and positional information, and show how to incorporate higher-order topological information, such as rings in molecular graphs. Our approach achieves state-of-the-art results on the ZINC benchmark dataset, while providing a flexible framework for encoding graph structure and incorporating higher-order topology.

Keywords

Cite

@article{arxiv.2304.10933,
  title  = {Self-Attention in Colors: Another Take on Encoding Graph Structure in Transformers},
  author = {Romain Menegaux and Emmanuel Jehanno and Margot Selosse and Julien Mairal},
  journal= {arXiv preprint arXiv:2304.10933},
  year   = {2023}
}
R2 v1 2026-06-28T10:13:39.547Z