Transfer learning for atomistic simulations using GNNs and kernel mean embeddings
This work addresses the computational cost of generating reference data for atomistic simulations, particularly in catalytic processes, and is incremental as it builds on pre-trained GNNs and kernel methods.
The paper tackles the challenge of generating large training datasets for accurate interatomic potentials in atomistic simulations by proposing a transfer learning algorithm that combines graph neural networks (GNNs) with kernel mean embeddings, showing excellent generalization and transferability performance and improving on existing methods.
Interatomic potentials learned using machine learning methods have been successfully applied to atomistic simulations. However, accurate models require large training datasets, while generating reference calculations is computationally demanding. To bypass this difficulty, we propose a transfer learning algorithm that leverages the ability of graph neural networks (GNNs) to represent chemical environments together with kernel mean embeddings. We extract a feature map from GNNs pre-trained on the OC20 dataset and use it to learn the potential energy surface from system-specific datasets of catalytic processes. Our method is further enhanced by incorporating into the kernel the chemical species information, resulting in improved performance and interpretability. We test our approach on a series of realistic datasets of increasing complexity, showing excellent generalization and transferability performance, and improving on methods that rely on GNNs or ridge regression alone, as well as similar fine-tuning approaches.