AI Paper Atlas

Research map / Graph learning / Graph self-supervised learning

Molecular Contrastive Learning with Chemical Element Knowledge Graph

Yin Fang, Qiang Zhang, Haihong Yang, Xiang Zhuang, Shumin Deng, Wen Zhang, Ming Qin, Zhuo Chen, Xiaohui Fan, Huajun Chen

AAAI 2022 · Graph learning › Graph self-supervised learning · approach: Molecular graph pretraining · 102 citations (OpenAlex)

This paper tried a common change: Add a contrastive loss

Add contrastive learning to KCL (GCN) · MoleculeNet classification tasks · ROC-AUC: 0.766 -> 0.852 Improved

What 1,317 accepted papers report for this change

See this change in detailMatched by the wording of each paper's own result, read by AI from the full text. Papers report gains more often than failures, so this shows what gets published, not the chance a change works.

Read the paper

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 Chemical Element KG-guided augmentation and KMPNN knowledge-aware encodingKCL for molecular property prediction · MoleculeNet fine-tune benchmarksROC-AUC and RMSE: 7.1% overall relative improvementImproved
Add contrastive learning to KCL (GCN) (ablation)KCL (GCN) · MoleculeNet classification tasksROC-AUC: 0.766 -> 0.852Improved
Add contrastive learning to KCL (GCN) (ablation)KCL (GCN) · MoleculeNet regression tasksRMSE: 1.984 -> 0.718Improved
Add contrastive learning to KCL (KMPNN) (ablation)KCL (KMPNN) · MoleculeNet classification tasksROC-AUC: 0.806 -> 0.852Improved
Add contrastive learning to KCL (KMPNN) (ablation)KCL (KMPNN) · MoleculeNet regression tasksRMSE: 1.531 -> 0.765Improved
Initialize attribute and relation features with KG embeddings rather than random initialization (ablation)KCL (GCN) · MoleculeNet classification tasksROC-AUC: 0.843 -> 0.852Improved
Use hard negative sampling rather than random negative sampling (ablation)KCL (GCN) · MoleculeNet classification tasksROC-AUC: 0.845 -> 0.852Improved
Initialize attribute and relation features with KG embeddings rather than random initialization (ablation)KCL (KMPNN) · MoleculeNet classification tasksROC-AUC: 0.845 -> 0.852Improved
Use hard negative sampling rather than random negative sampling (ablation)KCL (KMPNN) · MoleculeNet classification tasksROC-AUC: 0.847 -> 0.852Improved
Replace GCN with GIN or GAT on the original molecular graph (ablation)KCL · MoleculeNet downstream tasksROC-AUC: 0.849; 0.850; 0.852No clear effect
Replace KMPNN with RGCN or MPNN on augmented graphs (ablation)KCL · MoleculeNet classification tasksROC-AUC: 0.852 -> 0.831; 0.833Hurt
Use KMPNN instead of GCN for the augmented graph under linear evaluation (ablation)KCL · ToxCastROC-AUC: 0.696 -> 0.709No clear effect

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