An exceedingly large number of scientific and engineering fields are confronted with the need for computer simulations to study complex, real world phenomena or solve challenging design problems. When dealing with such computationally expensive simulation codes or process measurement data, surrogate modelling methods are firmly established as facilitators for design space exploration, sensitivity analysis, uncertainty quantification, optimization, visualization,etc.
After an introduction to surrogate modelling, this talk will explore surrogate-based optimization in more detail. First, Bayesian optimization based on the popular Gaussian process model will be introduced and how the same concept can be applied to solve other common engineering tasks. Secondly, to illustrate this further, a Bayesian optimization algorithm is presented that find all designs that satisfy the design requirements with a minimal number of expensive simulations.
Dr. Ivo Couckuyt is a post-doctoral researcher in the Internet and Data science Lab (IDLab) of Ghent University, Belgium. His research interests include surrogate modelling, surrogate-based optimization, and data-efficient machine learning applied to various time-consuming engineering problems, resulting in more than 100 peer-reviewed publications. He is the lead developer of the surrogate modelling toolbox and the Bayesian optimization toolbox GPFlowOpt, which have been used for solving many real-life applications in collaboration with industry and academia.
This event was hosted by the NAFEMS Optimisation Working Group (OWG). The OWG has formed an online Community to help disseminate best practice and encourage the adoption of optimisation methods and technology. You can discuss this and other presentations on the Optimisation Community Forum. For more information and to get involved go to the Optimisation Community webpage.
|Date||29th September 2020|