Gradient-Based Training and Pruning of Radial Basis Function Networks with an Application in Materials Physics
This work addresses the need for interpretable models in materials physics, offering incremental improvements through novel pruning criteria for continuous and binary data.
The authors tackled the challenge of interpretable and robust machine learning in physics by proposing a fully gradient-based technique for training and pruning radial basis function networks, resulting in compact models that provide insights into atomic configurations for atom-level migration processes in materials.
Many applications, especially in physics and other sciences, call for easily interpretable and robust machine learning techniques. We propose a fully gradient-based technique for training radial basis function networks with an efficient and scalable open-source implementation. We derive novel closed-form optimization criteria for pruning the models for continuous as well as binary data which arise in a challenging real-world material physics problem. The pruned models are optimized to provide compact and interpretable versions of larger models based on informed assumptions about the data distribution. Visualizations of the pruned models provide insight into the atomic configurations that determine atom-level migration processes in solid matter; these results may inform future research on designing more suitable descriptors for use with machine learning algorithms.