Accelerating PDE-constrained Inverse Solutions with Deep Learning and Reduced Order Models

Improving reduced order models by better characterizing their error using bounded neural network discrepancy models to accelerate many query applications including sampling-based Bayesian inference methods, and maximum a posteriori estimation.


Solving Forward and Inverse Problems with Model-Aware Autoencoders

UQ-VAE: a flexible, adaptive, hybrid data/model-informed framework for training neural networks to model the posterior distribution of parameters of interest.


Physics-informed convolutional autoencoders to compress time-stepping data


Recurrent Neural Networks to predict ocean storms given microseismic measurements

github / linkedin / twitter / email