AI Paper Atlas

Research map / Graph learning / Graph self-supervised learning

Fragment-based Pretraining and Finetuning on Molecular Graphs

Kha-Dinh Luong, Ambuj Singh

NeurIPS 2023 · Graph learning › Graph self-supervised learning · approach: Molecular graph pretraining

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
Add fragment encoder in downstream finetuning (C -> CF) (ablation)GraphFP contrastive · MoleculeNet 8 tasksaverage ROC-AUC: 72.81 -> 73.73Improved
Add predictive pretraining to contrastive and fragment encoder (CF -> CPF) (ablation)GraphFP · MoleculeNet 8 tasksaverage ROC-AUC: 73.73 -> 74.01Improved
Add predictive pretraining to contrastive (C -> CP) (ablation)GraphFP · MoleculeNet 8 tasksaverage rank: 5.50 -> 5.19Improved
Add predictive pretraining to contrastive (C -> CP) (ablation)GraphFP · MoleculeNet 8 tasksaverage ROC-AUC: 72.81 -> 72.59No clear effect
Use predictive pretraining alone instead of contrastive (ablation)GraphFP · MoleculeNet 8 tasksaverage ROC-AUC: 72.81 -> 69.05Hurt
Use JT-style smaller fragmentation and add fragment encoder (JT-C -> JT-CF) (ablation)GraphFP-JT · MoleculeNet 8 tasksaverage ROC-AUC: 72.66 -> 70.21Hurt
Increase fragment vocabulary size from 800 to 1600 (ablation)GraphFP-C · ClinToxROC-AUC: 78.6 -> 76.4Hurt
Increase fragment vocabulary size from 800 to 3200 (ablation)GraphFP-C · ClinToxROC-AUC: 78.6 -> 75.5Hurt
Increase fragment vocabulary size from 800 to 3200 (ablation)GraphFP-C · HIVROC-AUC: 76.3 -> 75.1Hurt
Increase fragment vocabulary size from 800 to 1600 (ablation)GraphFP-C · SIDERROC-AUC: 61.4 -> 60.4Hurt
Use alpha=0.3 instead of 0.1 in CP loss weighting (ablation)GraphFP-CP · MoleculeNet 8 tasksaverage rank: 2.88 -> 2.13Improved
Use alpha=0.1 instead of 0.5 in CPF loss weighting (ablation)GraphFP-CPF · MoleculeNet 8 tasksaverage rank: 3.00 -> 1.63Improved

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