Hierarchical Inter-Message Passing for Learning on Molecular Graphs

TL;DR

This paper introduces a hierarchical neural message passing architecture that leverages both molecular graphs and junction trees to improve molecular property prediction, overcoming classical GNN limitations.

Contribution

The novel hierarchical message passing model integrates graph and junction tree representations for enhanced molecular learning capabilities.

Findings

Improved performance on ZINC dataset.

Effective detection of cycles in molecular graphs.

Efficient training compared to classical GNNs.

Abstract

We present a hierarchical neural message passing architecture for learning on molecular graphs. Our model takes in two complementary graph representations: the raw molecular graph representation and its associated junction tree, where nodes represent meaningful clusters in the original graph, e.g., rings or bridged compounds. We then proceed to learn a molecule's representation by passing messages inside each graph, and exchange messages between the two representations using a coarse-to-fine and fine-to-coarse information flow. Our method is able to overcome some of the restrictions known from classical GNNs, like detecting cycles, while still being very efficient to train. We validate its performance on the ZINC dataset and datasets stemming from the MoleculeNet benchmark collection.