Prediction rigidities for data-driven chemistry

Abstract

The widespread application of machine learning (ML) to the chemical sciences is making it very important to understand how the ML models learn to correlate chemical structures with their properties, and what can be done to improve the training efficiency whilst guaranteeing interpretability and transferability. In this work, we demonstrate the wide utility of prediction rigidities, a family of metrics derived from the loss function, in understanding the robustness of ML model predictions. We show that the prediction rigidities allow the assessment of the model not only at the global level, but also on the local or the component-wise level at which the intermediate (e.g. atomic, body-ordered, or range-separated) predictions are made. We leverage these metrics to understand the learning behavior of different ML models, and to guide efficient dataset construction for model training. We finally implement the formalism for a ML model targeting a coarse-grained system to demonstrate the applicability of the prediction rigidities to an even broader class of atomistic modeling problems.

arXiv comment15 pages, 9 figures, submitted to Faraday Discussions: "Data-driven discovery in the chemical sciences", 2024

References in corpus (33)

• Gaussian Approximation Potentials: the accuracy of quantum mechanics, without the electrons
• On representing chemical environments
• PhysNet: A Neural Network for Predicting Energies, Forces, Dipole Moments and Partial Charges
• A Universal Graph Deep Learning Interatomic Potential for the Periodic Table
• Machine-learning based interatomic potential for amorphous carbon
• Nearsightedness of Electronic Matter
• A Fourth-Generation High-Dimensional Neural Network Potential with Accurate Electrostatics Including Non-local Charge Transfer
• Materials Cloud, a platform for open computational science
• Wide Neural Networks of Any Depth Evolve as Linear Models Under Gradient Descent
• A foundation model for atomistic materials chemistry
• Incorporating long-range physics in atomic-scale machine learning
• Data-driven learning of total and local energies in elemental boron
• Perspective on integrating machine learning into computational chemistry and materials science
• DeePCG: constructing coarse-grained models via deep neural networks
• Atom-Density Representations for Machine Learning
• Fast and Accurate Uncertainty Estimation in Chemical Machine Learning
• Transferable atomic multipole machine learning models for small organic molecules
• MatterSim: A Deep Learning Atomistic Model Across Elements, Temperatures and Pressures
• The Design Space of E(3)-Equivariant Atom-Centered Interatomic Potentials
• Uncertainty estimation for molecular dynamics and sampling
• Differentiable sampling of molecular geometries with uncertainty-based adversarial attacks
• Modelling atomic and nanoscale structure in the silicon-oxygen system through active machine learning
• Fast Uncertainty Estimates in Deep Learning Interatomic Potentials
• Heat transport in liquid water from first-principles and deep-neural-network simulations
• Uncertainty-biased molecular dynamics for learning uniformly accurate interatomic potentials
• Temperature- and vacancy-concentration-dependence of heat transport in Li${}_3$ClO from multi-method numerical simulations
• Robustness of Local Predictions in Atomistic Machine Learning Models
• Exploring the configurational space of amorphous graphene with machine-learned atomic energies
• A smooth basis for atomistic machine learning
• Wigner kernels: body-ordered equivariant machine learning without a basis
• Variational Bayesian Last Layers
• A prediction rigidity formalism for low-cost uncertainties in trained neural networks
• Data-efficient fine-tuning of foundational models for first-principles quality sublimation enthalpies

Cited by in corpus (5)

• Fast and flexible long-range models for atomistic machine learning
• Metatensor and metatomic: foundational libraries for interoperable atomistic machine learning
• Resolving the Body-Order Paradox of Machine Learning Interatomic Potentials
• Convergence of Body-Orders in Linear Atomic Cluster Expansions
• Assessing zero-shot generalisation behaviour in graph-neural-network interatomic potentials