Speaker
Description
Broadly speaking, Bayesian optimisation methods for a single objective function (without constraints) proceed by (i) assuming a prior for the unknown function f (ii) selecting new points x at which to evaluate f according to some infill criterion that maximises an acquisition function; and (iii) updating an estimate of the function optimum, and its location, using the updated posterior for f. The most common prior for f is a Gaussian process (GP).
Optimisation under uncertainty is important in many areas of research. Uncertainty can come from various sources, including uncertain inputs, model uncertainty, code uncertainty and others. Multi-objective optimisation under uncertainty is a powerful tool and a big area of research.
In this talk, I will give an overview of Bayesian optimisation and talk about a few extensions to the emulation-based optimisation methodology called expected quantile improvement (EQI) to a two-objective optimisation case. We demonstrate how this multi-objective optimisation technique handles uncertainty and finds optimal solutions under high levels of uncertainty.
| Classification | Mainly methodology |
|---|---|
| Keywords | DOE, Bayesian, Optimisation |