Path-Augmented Graph Transformer Network

Benson Chen, Regina Barzilay, Tommi Jaakkola
5/29/2019
cs.LGstat.ML

Official code: benatorc/PA-Graph-Transformer (44 stars, MIT, last commit 2023-06-16).

Abstract

Much of the recent work on learning molecular representations has been based on Graph Convolution Networks (GCN). These models rely on local aggregation operations and can therefore miss higher-order graph properties. To remedy this, we propose Path-Augmented Graph Transformer Networks (PAGTN) that are explicitly built on longer-range dependencies in graph-structured data. Specifically, we use path features in molecular graphs to create global attention layers. We compare our PAGTN model against the GCN model and show that our model consistently outperforms GCNs on molecular property prediction datasets including quantum chemistry (QM7, QM8, QM9), physical chemistry (ESOL, Lipophilictiy) and biochemistry (BACE, BBBP).

View on arXivView PDF

Code Implementations(1)

benatorc/PA-Graph-TransformerOfficial100%
MIT
4412Shell, PythonMay 28, 20193 years agoMIT

Similar Papers

FunQG: Molecular Representation Learning Via Quotient Graphs

88% match1 repo122022

Semi-Supervised Junction Tree Variational Autoencoder for Molecular Property Prediction

87% match2 repos122022

Neural Message Passing for Quantum Chemistry

87% match1 repo2392017

Molecular Property Prediction Based on Graph Structure Learning

86% match2023

RingFormer: A Ring-Enhanced Graph Transformer for Organic Solar Cell Property Prediction

86% match1 repo72024

Synergistic Fusion of Graph and Transformer Features for Enhanced Molecular Property Prediction

86% match2023

Cite this paper

@article{chen2019pathaugmented,
  title  = {Path-Augmented Graph Transformer Network},
  author = {Benson Chen and Regina Barzilay and Tommi Jaakkola},
  year   = {2019},
  eprint = {1905.12712},
  archivePrefix = {arXiv},
  url    = {https://arxiv.org/abs/1905.12712}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/1905.12712.svg)](https://codeofpaper.com/paper/1905.12712)

Discussion