Relational Graph Attention Networks
By Dan Busbridge, Dane Sherburn, Pietro Cavallo, Nils Y. Hammerla. 2018
247 citations recorded by the source corpus.
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.
Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Statistical and Nonlinear Physics
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Uri Alon
- Charles Sutton
- Eran Yahav
- Yoshua Bengio
- Miltiadis Allamanis
- Percy Liang
- Quoc V. Le
- Shaked Brody
In Computer Vision and Pattern Recognition
- Yoshua Bengio
- Luc Van Gool
- Ming-Hsuan Yang
- Dacheng Tao
- Jitendra Malik
- Trevor Darrell
- Bernt Schiele
- Xiaogang Wang
In Statistical and Nonlinear Physics
- Daniel Tarlow
- Shaked Brody
- Eran Yahav
- Hugo Larochelle
- Katsuhiko Ishiguro
- Masanori Koyama
- Shin-ichi Maeda
- Charles Sutton
Citations in the corpus
- Graph Warp Module: an Auxiliary Module for Boosting the Power of Graph Neural Networks (2019, methodology): “In our formulation, the attention similarity measure a ( · ) is defined by the general attention in [16] while the original GAT [28] and the RGAT [2] employed a simpler concat attention: B Experiments Details”
- How Attentive are Graph Attention Networks? (2021, methodology): “…(Zhang et al., 2018; Thekumparampil et al., 2018; Gao and Ji, 2019; Lukovnikov and Fischer, 2021; Shi et al., 2020; Dwivedi and Bresson, 2020; Busbridge et al., 2019; Rong et al., 2020a; Veličković et al., 2020), and Lee et al. (2018) conducted an extensive survey of attention types in…”
- Graph Warp Module: an Auxiliary Module for Boosting the Power of Graph Neural Networks in Molecular Graph Analysis (2019, methodology): “1, thin dashed lines), and this seems to happen to numerous GNN architectures that are used in applications today [7, 17, 14, 31, 4].”
- Learning to Execute Programs with Instruction Pointer Attention Graph Neural Networks (2020, methodology): “It is extended by R-GAT [6] to support distinct edge types.”