Tid: 28 november, kl 13-14 Plats: B705

Abstract

Gaussian processes are widely used for flexible modeling of quite complex phenomena in the machine learning field. The talk gives an introduction to Gaussian process regression and classification with a view toward applications. I will also briefly discuss how Gaussian processes can be used to recast numerical optimization and integration problems as Bayesian inference problems. The methods will be illustrated in some recent applications in neuroimaging, robotics, finance and transportation.