Information fusion in multi-task Gaussian processes

TL;DR

This paper demonstrates that multi-task Gaussian processes effectively fuse heterogeneous information sources, improving estimates in geological resource modeling by considering correlations among multiple quantities, based on large-scale sensor data.

Contribution

It empirically shows the benefits of multi-task Gaussian processes for integrating diverse data sources in geological modeling, a novel application in this context.

Findings

Information fusion improves modeling accuracy

Multi-task Gaussian processes outperform single-task models

Effective on large-scale real sensor data

Abstract

This paper evaluates heterogeneous information fusion using multi-task Gaussian processes in the context of geological resource modeling. Specifically, it empirically demonstrates that information integration across heterogeneous information sources leads to superior estimates of all the quantities being modeled, compared to modeling them individually. Multi-task Gaussian processes provide a powerful approach for simultaneous modeling of multiple quantities of interest while taking correlations between these quantities into consideration. Experiments are performed on large scale real sensor data.