Received wisdom would have us believe that inductive biases imposed on learning models can ultimately hold back progress, and that we may be better off applying more compute and gathering more data. And yet, physical systems such as molecules and materials obey well understood laws that cannot be violated. In this talk, I will outline how Euclidean symmetry-equivariant Neural Networks (E(3)NNs) provide a powerful counterexample where incorporating the symmetries of 3D space allows us to achieve state-of-the-art performance on a variety of learning tasks focused on predicting properties of atomistic systems. Beyond the exceptional accuracy of these models, they also exhibit excellent data-efficiency, a particularly important property when labels typically come from time-consuming experiments or quantum-mechanical calculations.
I will give a few examples of how we have developed E(3)NNs and other symmetry-aware models to predict materials properties such as interaction potentials, Born effective charges, Raman tensors, NMR shielding tensors and self-consistent Hubbard corrections to significantly improve the quality of density functional theory calculations for transition elements at almost no computational cost.
Laboratory for Materials Simulations (LMS)