Neural network potentials for chemistry: concepts, applications and prospects

Abstract

Artificial Neural Networks (NN) are already heavily involved in methods and applications for frequent tasks in the field of computational chemistry such as representation of potential energy surfaces (PES) and...

Machine-learning based prediction of small molecule–surface interaction potentials

“…Accurately capturing all the underlying effects in a simple analytical model is not feasible 12 and thus we turn to a machine learning (ML) approach for prediction. Many groups have already approached the problem of the binding of ligands to specic targets using ML techniques, as well as the more general cases of the prediction of PMFs, potentials for complex systems or indeed entire forceelds, 4,[13][14][15][16][17][18][19] suggesting this is a suitable methodology to apply to the prediction of molecule-surface adsorption.…”

Generalizing property prediction of ionic liquids from limited labeled data: a one-stop framework empowered by transfer learning

“…Nevertheless, both the two schemes are prone to the inherent weakness of limited predictive power and/or insufficient accuracy. 8 Another computational method for IL property prediction is the quantitative structure-property relationship (QSPR) approach, wherein a property of interest is correlated quantitatively with certain descriptors of involved molecules 9,10 (for which machine learning methods have recently gained popularity [11][12][13][14][15][16][17] ). Notably, the availability of IL property databases like ILThermo 18 has stimulated the use of ML methods for modeling IL properties, wherein diverse types of molecular descriptors were used as IL representation.…”

Lifelong Machine Learning Potentials

“…By contrast, machine learning potentials (MLPs) − can preserve the high accuracy of electronic structure methods but at low computational cost comparable to that of force fields. MLPs are not based on physical approximations but rely on very flexible mathematical expressions to obtain an analytical potential energy surface.…”