CMStatistics 2017: Start Registration
View Submission - CMStatistics
Title: Modelling covariance structure of multi-dimensional function-valued processes using convolved Gaussian processes Authors:  Evandro Konzen - Newcastle University (United Kingdom) [presenting]
Jianqing Shi - Newcastle University (United Kingdom)
Abstract: In order to consider the cross-covariance structure among several multi-dimensional functions, we use convolved Gaussian processes ensuring that the covariance matrix for the multivariate functional response is positive definite. We discuss implementation issues and show examples based on simulated data. An application to human fertility data is also illustrated.