Materials Data Analytics and Machine Learning
Uncertainty Quantification for Material Response
With the future of materials design heading towards a heavy reliability on models to simulate and predict material behavior, quantifying the confidence, or equivalently the uncertainty, in these model predictions must be achieved in a principled way. The goal of this project is to adapt and apply advanced statistical tools, namely in the Bayesian paradigm, to establish the uncertainty in material models. This approach allows various sources of uncertainty to be accounted for, including but not limited to model discrepancy, material response, experimental variability and measurement noise in order determine the range of most likely material responses from a deterministic model.