Graph Neural Network Architecture Search for Molecular Property Prediction

Shengli Jiang,Prasanna Balaprakash

DOI: https://doi.org/10.48550/arXiv.2008.12187

2020-08-27

Abstract:Predicting the properties of a molecule from its structure is a challenging task. Recently, deep learning methods have improved the state of the art for this task because of their ability to learn useful features from the given data. By treating molecule structure as graphs, where atoms and bonds are modeled as nodes and edges, graph neural networks (GNNs) have been widely used to predict molecular properties. However, the design and development of GNNs for a given data set rely on labor-intensive design and tuning of the network architectures. Neural architecture search (NAS) is a promising approach to discover high-performing neural network architectures automatically. To that end, we develop an NAS approach to automate the design and development of GNNs for molecular property prediction. Specifically, we focus on automated development of message-passing neural networks (MPNNs) to predict the molecular properties of small molecules in quantum mechanics and physical chemistry data sets from the MoleculeNet benchmark. We demonstrate the superiority of the automatically discovered MPNNs by comparing them with manually designed GNNs from the MoleculeNet benchmark. We study the relative importance of the choices in the MPNN search space, demonstrating that customizing the architecture is critical to enhancing performance in molecular property prediction and that the proposed approach can perform customization automatically with minimal manual effort.

Machine Learning,Biomolecules

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to automatically design and develop efficient graph neural network (GNN) architectures, especially message - passing neural networks (MPNN), in molecular property prediction. Traditional methods require a great deal of manual labor when designing and tuning GNN architectures, which limits their performance improvement on different datasets and wide - scale application. To address this issue, the paper proposes a method based on neural architecture search (NAS), aiming to automatically discover high - performance MPNN architectures suitable for specific tasks, thereby reducing the complexity of manual design and improving the prediction accuracy of the model.

Specifically, the paper focuses on using NAS technology to automatically generate MPNN architectures for small - molecule quantum mechanics and physicochemical property prediction. By comparing with manually - designed GNNs, it demonstrates the superiority of the automatically - discovered MPNN architectures on multiple benchmark datasets, and explores the importance of different choices in the MPNN search space, emphasizing the crucial role of customized architectures in enhancing molecular property prediction performance.