Relational Graph Attention Networks

TL;DR

This paper explores Relational Graph Attention Networks, comparing their performance to related models, and finds they are generally less effective but may have specific benefits for molecular property modeling.

Contribution

It provides a comprehensive evaluation of Relational Graph Attention Networks and their spectral counterparts, highlighting their limitations and potential in certain applications.

Findings

Relational Graph Attention Networks perform worse than expected.

Some configurations marginally benefit molecular property modeling.

Insights and future directions are proposed for improving these models.

Abstract

We investigate Relational Graph Attention Networks, a class of models that extends non-relational graph attention mechanisms to incorporate relational information, opening up these methods to a wider variety of problems. A thorough evaluation of these models is performed, and comparisons are made against established benchmarks. To provide a meaningful comparison, we retrain Relational Graph Convolutional Networks, the spectral counterpart of Relational Graph Attention Networks, and evaluate them under the same conditions. We find that Relational Graph Attention Networks perform worse than anticipated, although some configurations are marginally beneficial for modelling molecular properties. We provide insights as to why this may be, and suggest both modifications to evaluation strategies, as well as directions to investigate for future work.