Predicting fault slip via transfer learning

By teaching machine learning models with earthquake fault numerical simulations laboratory fault slip is predictable. Training the model further with a snippet of laboratory data improves predictions suggesting an approach to probing faults in Earth.