GPS++: Reviving the Art of Message Passing for Molecular Property Prediction
Abstract
We present GPS++, a hybrid Message Passing Neural Network / Graph Transformer model for molecular property prediction. Our model integrates a well-tuned local message passing component and biased global attention with other key ideas from prior literature to achieve state-of-the-art results on large-scale molecular dataset PCQM4Mv2. Through a thorough ablation study we highlight the impact of individual components and find that nearly all of the model’s performance can be maintained without any use of global self-attention, showing that message passing is still a competitive approach for 3D molecular property prediction despite the recent dominance of graph transformers. We also find that our approach is significantly more accurate than prior art when 3D positional information is not available.
Cite
Text
Masters et al. "GPS++: Reviving the Art of Message Passing for Molecular Property Prediction." Transactions on Machine Learning Research, 2023.Markdown
[Masters et al. "GPS++: Reviving the Art of Message Passing for Molecular Property Prediction." Transactions on Machine Learning Research, 2023.](https://mlanthology.org/tmlr/2023/masters2023tmlr-gps/)BibTeX
@article{masters2023tmlr-gps,
title = {{GPS++: Reviving the Art of Message Passing for Molecular Property Prediction}},
author = {Masters, Dominic and Dean, Josef and Klaeser, Kerstin and Li, Zhiyi and Maddrell-Mander, Samuel and Sanders, Adam and Helal, Hatem and Beker, Deniz and Fitzgibbon, Andrew W and Huang, Shenyang and Rampášek, Ladislav and Beaini, Dominique},
journal = {Transactions on Machine Learning Research},
year = {2023},
url = {https://mlanthology.org/tmlr/2023/masters2023tmlr-gps/}
}