Adam Subel is the lead author in this article , which builds a framework to understand and implement transfer learning, a powerful machine learning technique for enabling generalization between datasets from different configurations of a system or of varying quality/quantity, in multiscale physical systems.