Hybrid localized graph kernel for machine learning energy-related properties of molecules and solids

TL;DR

This paper introduces a hybrid localized graph kernel that combines chemical pattern recognition and local geometric details to improve energy property predictions of molecules and solids in machine learning models.

Contribution

It presents a novel graph-based descriptor with a hybrid kernel for regression tasks, enhancing prediction accuracy over existing methods.

Findings

Outperforms SOAP and Coulomb matrix methods in energy predictions.

Effective on QM7 and BA10 datasets for molecules and solids.

Demonstrates improved accuracy in energy-related property predictions.

Abstract

Nowadays, the coupling of electronic structure and machine learning techniques serves as a powerful tool to predict chemical and physical properties of a broad range of systems. With the aim of improving the accuracy of predictions, a large number of representations for molecules and solids for machine learning applications has been developed. In this work we propose a novel descriptor based on the notion of molecular graph. While graphs are largely employed in classification problems in cheminformatics or bioinformatics, they are not often used in regression problem, especially of energy-related properties. Our method is based on a local decomposition of atomic environments and on the hybridization of two kernel functions: a graph kernel contribution that describes the chemical pattern and a Coulomb label contribution that 1encodes finer details of the local geometry. The accuracy of this new kernel method in energy predictions of molecular and condensed phase systems is demonstrated by considering the popular QM7 and BA10 datasets. These examples show that the hybrid localized graph kernel outperforms traditional approaches such as, for example, the smooth overlap of atomic positions (SOAP) and the Coulomb matrices.