The National Science Foundation (NSF) Division of Mathematical Sciences awarded a $220,000 Project Grant to the International Computer Science Institute (ICSI), a non-profit research organization, under the Mathematical and Physical Sciences program (CFDA 47.049). The project aims to develop resilient and reliable deep learning methods for forecasting complex spatiotemporal ground motion data, with applications in seismology, earth sciences, and other domains. Key technical objectives include learning continuous dynamics, modeling multiscale structures in space and time, and improving the robustness of neural networks to input data perturbations. The project will deliver neural network architectures for learning robust latent space embeddings and enhanced forecasting capabilities, accounting for uncertainties in ground motion data. Both observed and simulated data will be used to demonstrate the advantages of the proposed methods, which combine dynamical systems theory, seismology, and deep learning.
Generated 4/30/24, 2:17 PM