Managing Uncertainty in Complex Models (2006-2012)
The Managing Uncertainty in Complex Models (MUCM) project involved developing methodology for emulation, sensitivity analysis, experimental design, calibration and history matching. You can read about all this in the MUCM Toolkit
Calibration and analysis of complex models: methodological development and application to explore the impact of HAART in Africa (2012-2016)
This project, led by Richard White at LSHTM, has involved the use of emulation and history matching methods for calibrating stochastic infectious disease models to observed data. A tutorial paper is available here.
Simulation Tools for Automated and Robust Manufacturing (2013-2016)
This project is in collaboration with the Advanced Manufacturing Research Centre with Boeing. The aim is to develop automated machining processes that are robust to variation in material properties. One theme within the project is to explore the use of (computationally expensive) finite element models together with emulation and calibration methods to infer material properties indirectly from sensor data, and make process adjustments. Keith Harris is working as the RA on this project.
Knowledge Transfer Partnership with HR Wallingford (2014-17)
HR Wallingford use numerical and physical modelling in relation to civil engineering and environmental hydraulics. The aim of this KTP is to implement emulator and other UQ methods within their modelling framework. Sajni Malde is working as the KTP associate on this project.
Uncertainty Quantification in Prospective and Predictive Patient Specific Cardiac Models (2017-2021)
This project is led by Richard Clayton (PI), with Jeremy Oakley and Richard Wilkinson. We are looking at how patient specific computational models can be used to make prospective predictions to guide procedures and inform uncertain clinical decisions using complex cardiac simulation models.
Machine Learning for Aerosol Lifetime Prediction (2017-2018)
Industrial support from SC Johnson, led by Richard Wilkinson, with Tony Ryan and Alan Saul. In this project, we are looking at using Gaussian processes to improve the time-to-market of new aerosol products.