Problem
Research questions and friction points this paper is trying to address.
Few-shot molecular property prediction
Label agreement
Property dependence
Class marginals
CalibHyper: Chance-Corrected Relational Hypergraphs for Few-Shot Molecular Property Prediction
🤖 AI Summary
This study addresses the challenges of label scarcity, sensitivity of label consistency to class distributions, and difficulty in capturing attribute dependencies in molecular property prediction. We propose an opportunity-corrected relational hypergraph method based on joint label distributions. Specifically, this approach removes the independence baseline via subtraction and shrinks the residual, employing a permutation-equivariant head to estimate residuals for adaptively selecting auxiliary attributes and assigning hyperedge weights. By integrating relational hypergraphs with permutation-equivariant networks, it directly models inter-attribute dependencies. Extensive experiments across thirteen datasets from five benchmarks demonstrate that the proposed method achieves ROC-AUC performance comparable to state-of-the-art results under both 1-shot and 10-shot settings.
📝 Abstract
Molecular property prediction is central to drug development and materials discovery, but experiments are costly and labeled data are scarce. Context-aware methods use auxiliary assay labels to support few-shot prediction, and recent work supervises property relations with label agreement. However, label agreement is sensitive to class marginals and does not directly capture dependence between properties. We propose CalibHyper, a chance-corrected relational hypergraph method based on the joint label distribution. CalibHyper subtracts an independence baseline from the ordered four-state label distribution and shrinks the residual according to the number of joint observations. A swap-equivariant relation head estimates these residuals, which choose the auxiliary properties for each molecule and set the sign and weight of their hyperedge messages. On thirteen datasets from five benchmarks, in both 1-shot and 10-shot settings, CalibHyper and its ablation settings achieve ROC-AUC competitive with the strongest reported results.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Chance-Corrected Relational Hypergraphs
Few-Shot Molecular Property Prediction
Joint Label Distribution
Swap-Equivariant Relation Head
Context-Aware Learning
🔎 Similar Papers
No similar papers found.
