Communication: Understanding molecular representations in machine learning: The role of uniqueness and target similarity

The predictive accuracy of Machine Learning (ML) models of molecular properties depends on the choice of the molecular representation. Inspired by the postulates of quantum mechanics, we introduce a hierarchy of representations which meet uniqueness and target similarity criteria. To systematically...

Full description

Saved in:
Bibliographic Details
Published in:The Journal of chemical physics 2016-10, Vol.145 (16), p.161102-161102
Main Authors: Huang, Bing, von Lilienfeld, O. Anatole
Format: Article
Language:English
Subjects:
Accuracy
Artificial intelligence
Atomizing
Density functional theory
Dipole moments
Electron affinity
Enthalpy
Ionization potentials
Machine learning
Mathematical models
Model accuracy
Molecular orbitals
Molecular properties
Numerical prediction
Organic chemistry
Physics
Potential energy
Quantum mechanics
Representations
Similarity
Uniqueness
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The predictive accuracy of Machine Learning (ML) models of molecular properties depends on the choice of the molecular representation. Inspired by the postulates of quantum mechanics, we introduce a hierarchy of representations which meet uniqueness and target similarity criteria. To systematically control target similarity, we simply rely on interatomic many body expansions, as implemented in universal force-fields, including Bonding, Angular (BA), and higher order terms. Addition of higher order contributions systematically increases similarity to the true potential energy and predictive accuracy of the resulting ML models. We report numerical evidence for the performance of BAML models trained on molecular properties pre-calculated at electron-correlated and density functional theory level of theory for thousands of small organic molecules. Properties studied include enthalpies and free energies of atomization, heat capacity, zero-point vibrational energies, dipole-moment, polarizability, HOMO/LUMO energies and gap, ionization potential, electron affinity, and electronic excitations. After training, BAML predicts energies or electronic properties of out-of-sample molecules with unprecedented accuracy and speed.
ISSN:0021-9606
1089-7690
DOI:10.1063/1.4964627