Deep Learning Meets Earthquake Aftershocks

This chapter explores the innovative integration of deep learning techniques to model earthquake aftershocks, emphasizing the creation of a neural network that incorporates physical geology. The discussion highlights the challenges of data assembly and physical regularization while revealing surprising insights that simplify predictions based on a single physical quantity. Through a thoughtful examination of stress changes in the Earth's crust, the speakers illustrate the potential of machine learning to enhance our understanding of seismic activities.