AI Paper Atlas

Research map / Label-efficient and robust learning / OOD and anomaly detection

Identifying Semantic Component for Robust Molecular Property Prediction

Zijian Li, Zunhong Xu, Ruichu Cai, Zhenhui Yang, Yuguang Yan, Zhifeng Hao, Guangyi Chen, Kun Zhang

TPAMI 2025 · Label-efficient and robust learning › OOD and anomaly detection · approach: Robust fine-tuning for OOD generalization · 19 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
add semantic latent substructure regularization L_r (ablation)SCI · MoleculeNet/OGB/GOOD datasets (Figure ablation)ROC-AUCImproved
add sparsity regularization L_h on G_h (ablation)SCI · MoleculeNet/OGB/GOOD datasets (Figure ablation)ROC-AUCImproved
add sparsity regularization L_n on G_ir (ablation)SCI · MoleculeNet/OGB/GOOD datasets (Figure ablation)ROC-AUCImproved
add atom classifier for atom latent variable identification (ablation)SCI · MoleculeNet/OGB/GOOD datasets (Figure ablation)ROC-AUCImproved
set alpha around 1.0 (ablation)SCI · MoleculeNet Bace, Sider, ClinToxROC-AUCImproved
increase beta (stronger sparsity penalty on SR substructure) (ablation)SCI · MoleculeNet Bace, Sider, ClinToxROC-AUCHurt
use sparsity regularization with beta=2 instead of none (ablation)SCI · MoleculeNet Bace, Sider, ClinToxROC-AUCImproved
SCI versus strongest baseline OOD-GNNSCI · OGB MolfreesolvRMSE: 1.8100 -> 1.9484Hurt
SCI versus strongest baseline AttentiveFPSCI · MoleculeNet HIVROC-AUC: 0.7662 -> 0.7546Hurt
SCI versus GSAT on GOOD-HIV scaffoldGSAT · GOOD-HIV (scaffold split)ROC-AUC (%): 70.07 -> 72.15Improved
SCI versus best baseline on OGB MolsiderAttentiveFP · OGB Molsider (scaffold)ROC-AUC: 0.6919 -> 0.7152Improved

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