Weisfeiler-Lehman Embedding for Molecular Graph Neural Networks

Abstract

A graph neural network (GNN) is a good choice for predicting the chemical properties of molecules. Compared with other deep networks, however, the current performance of a GNN is limited owing to the "curse of depth." Inspired by long-established feature engineering in the field of chemistry, we expanded an atom representation using Weisfeiler-Lehman (WL) embedding, which is designed to capture local atomic patterns dominating the chemical properties of a molecule. In terms of representability, we show WL embedding can replace the first two layers of ReLU GNN -- a normal embedding and a hidden GNN layer -- with a smaller weight norm. We then demonstrate that WL embedding consistently improves the empirical performance over multiple GNN architectures and several molecular graph datasets.

arXiv commentReference Updated. An implementation example is included in Chainer Chemistry Ver 0.7.1: see https://github.com/chainer/chainer-chemistry

References in corpus (9)

• Inductive Representation Learning on Large Graphs
• Neural Message Passing for Quantum Chemistry
• Understanding deep learning requires rethinking generalization
• MolGAN: An implicit generative model for small molecular graphs
• Relational Graph Attention Networks
• TeaNet: universal neural network interatomic potential inspired by iterative electronic relaxations
• N-Gram Graph: Simple Unsupervised Representation for Graphs, with Applications to Molecules
• Graph Warp Module: an Auxiliary Module for Boosting the Power of Graph Neural Networks in Molecular Graph Analysis
• Graph Classification via Deep Learning with Virtual Nodes

Cited by in corpus (1)

• Boosting Heterogeneous Catalyst Discovery by Structurally Constrained Deep Learning Models