May 4, 2018 @ 10:00 am - 11:00 am
In this talk, we will present a new approach to a data-driven, learning-based framework for predicting outcomes of physical systems and for discovering hidden physics from noisy data. A key concept is the seamless fusion and integration of data of variable fidelity into the predictive models. First, we will present a Bayesian approach using Gaussian Process Regression, and subsequently a deep learning approach based on neural networks. Unlike other approaches that rely on big data, here we “learn” from small data by exploiting the information provided by the physical conservation laws, which are used to obtain informative priors or regularize the neural networks.
This work is supported by the DARPA EQUiPS program on Uncertainty Qualification and AFOSR.