Output details

Computationally Efficient Convolved Multiple Output Gaussian Processes

<24>Multiple output regression models capture the correlation between variables at given input locations. For example, in geostatistics, we might jointly model pH and heavy metal concentration. We can then use pH as a proxy for expensive heavy metal concentration measurements. Such correlations can be expressed as differential equations, often leading to computationally demanding convolutions for their solutions. In this paper we present the first practical approaches to speeding up these computations. The paper [GoogleScholar:12, Arxiv:39], led to an invite for a review paper (with Rosasco of MIT - DOI:10.1561/2200000036). JMLR has an IF 3.42.