English

Multi-resolution Multi-task Gaussian Processes

Machine Learning 2019-11-06 v2 Machine Learning

Abstract

We consider evidence integration from potentially dependent observation processes under varying spatio-temporal sampling resolutions and noise levels. We develop a multi-resolution multi-task (MRGP) framework while allowing for both inter-task and intra-task multi-resolution and multi-fidelity. We develop shallow Gaussian Process (GP) mixtures that approximate the difficult to estimate joint likelihood with a composite one and deep GP constructions that naturally handle biases in the mean. By doing so, we generalize and outperform state of the art GP compositions and offer information-theoretic corrections and efficient variational approximations. We demonstrate the competitiveness of MRGPs on synthetic settings and on the challenging problem of hyper-local estimation of air pollution levels across London from multiple sensing modalities operating at disparate spatio-temporal resolutions.

Keywords

Cite

@article{arxiv.1906.08344,
  title  = {Multi-resolution Multi-task Gaussian Processes},
  author = {Oliver Hamelijnck and Theodoros Damoulas and Kangrui Wang and Mark Girolami},
  journal= {arXiv preprint arXiv:1906.08344},
  year   = {2019}
}