AI Paper Atlas

Research map / Graph learning / Graph self-supervised learning

3D Infomax improves GNNs for Molecular Property Prediction

Hannes Stärk, Dominique Beaini, Gabriele Corso, Prudencio Tossou, Christian Dallago, Stephan Günnemann, Pietro Lió

ICML 2022 · Graph learning › Graph self-supervised learning · approach: Molecular graph pretraining · 108 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
3D Infomax pre-training (contrastive 2D-3D MI maximization)PNA pre-trained on QM9 subset · QM9 quantum propertiesMAE (average ~22% reduction): ~22% average MAE reductionImproved
3D Infomax pre-training vs GraphCLPNA · QM9 quantum properties (homo)MAE: 68.96 vs 79.57 (GraphCL)Improved
3D Infomax pre-training on GEOM-Drugs vs GraphCLPNA · GEOM-Drugs quantum properties (Gibbs, <E>)MAE: .1811 vs .1941 (Gibbs)Improved
Pre-training with 2D-only self-supervision (GraphCL node dropping) or property prediction (ProPred)PNA · QM9 homoMAE: ProPred 93.11 vs Rand Init 82.10Hurt
Distance prediction pre-training (DisPred) on GEOM-DrugsPNA · QM9 quantum propertiesMAE: e.g. mu 0.4626 vs 0.4133 worse; homo 80.58 vs 82.10 betterMixed
Using multiple conformers (sum over conformer similarities in contrastive loss) (ablation)3D Infomax with QMugs pre-training · QM9 homoMAE: 82.57 with single conformer vs 70.77 with threeImproved
Increasing number of GEOM-Drugs conformers beyond ~3 during pre-training (ablation)3D Infomax on GEOM-Drugs · GEOM-Drugs Gibbs and QM9 homoMAEHurt
Distance encoding gamma (sin/cos high-frequency features) in 3D network (ablation)Net3D · QM9 homo pre-trainingMAE: 68.96 (Net3D) vs 70.34 (w/o gamma)Improved
Net3D 3D network instead of SMP or EGNN (ablation)3D Infomax 3D network · QM9 homo pre-trainingMAE: 68.96 vs SMP 72.37 and EGNN 70.46Improved
3D Infomax on non-quantum OGB/MoleculeNet datasets (GEOM-Drugs pre-training)PNA · 10 non-quantum datasetsRMSE / ROC-AUC: improved 4 of 10 datasets; never decreased performanceMixed
3D Infomax pre-training with RDKit-generated conformers in SMP baselineSMP (3D GNN, fine-tuning with explicit RDKit 3D) · QM9 homoMAE: 82.51 vs Rand Init 82.10Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy