Swiss Equivariant Machine Learning Workshop

Machine learning (ML) based atomistic models are often faced with the challenge of learning physical or chemical properties that have well-defined transformations (equivariance) under translation, rotation,  and reflections of the corresponding atomic structures. The last few

years have seen significant progress on enhancing ML approaches by including symmetry preserving operations, which has led to noteworthy improvements in both the accuracy and data-efficiency of models predicting physical quantities.