PCEvo: Path-Consistent Molecular Representation via Virtual Evolutionary

TL;DR

PCEvo introduces a path-consistent molecular representation learning method that leverages virtual evolutionary paths to improve few-shot property prediction accuracy in molecular datasets.

Contribution

It proposes a novel path-consistency objective that enforces invariance across multiple virtual evolutionary paths, enhancing few-shot generalization in molecular property prediction.

Findings

Significant improvement in few-shot prediction accuracy on QM9 and MoleculeNet datasets.

Effective enforcement of prediction invariance across alternative molecular paths.

Demonstrated robustness in limited data scenarios.

Abstract

Molecular representation learning aims to learn vector embeddings that capture molecular structure and geometry, thereby enabling property prediction and downstream scientific applications. In many AI for science tasks, labeled data are expensive to obtain and therefore limited in availability. Under the few-shot setting, models trained with scarce supervision often learn brittle structure-property relationships, resulting in substantially higher prediction errors and reduced generalization to unseen molecules. To address this limitation, we propose PCEvo, a path-consistent representation method that learns from virtual paths through dynamic structural evolution. PCEvo enumerates multiple chemically feasible edit paths between retrieved similar molecular pairs under topological dependency constraints. It transforms the labels of the two molecules into stepwise supervision along each virtual evolutionary path. It introduces a path-consistency objective that enforces prediction invariance across alternative paths connecting the same two molecules. Comprehensive experiments on the QM9 and MoleculeNet datasets demonstrate that PCEvo substantially improves the few-shot generalization performance of baseline methods. The code is available at https://anonymous.4open.science/r/PCEvo-4BF2.