Prototype learning with structural-semantic alignment for interpretable molecular relational learning

Peiliang Zhang

doi:10.1016/j.knosys.2026.115460

Prototype learning with structural-semantic alignment for interpretable molecular relational learning

2026-02-02

0

PRE

AI

DOI：10.1016/j.knosys.2026.115460

Original

Original request for help

Save

Abstract

En 中文

• PSSA mitigates inconsistencies in MRL caused by conformation sensitivity. • ChemAlign integrates molecular information by structure-guided semantic learning. • RCProto enhances interpretability by prototype generation, fine-tuning, and matching. • PSSA outperforms comparative models in predictive performance and interpretability.

Keywords:

PSSA

ChemAlign

RCProto

prototype learning

molecular relational learning

Prototype learning with structural-semantic alignment for interpretable molecular relational learning

Abstract

Journal

Organization