Variational Inference for Gaussian Process Modulated Poisson Processes

Chris Lloyd, Tom Gunter, Michael Osborne, Stephen Roberts
Proceedings of the 32nd International Conference on Machine Learning, PMLR 37:1814-1822, 2015.

Abstract

We present the first fully variational Bayesian inference scheme for continuous Gaussian-process-modulated Poisson processes. Such point processes are used in a variety of domains, including neuroscience, geo-statistics and astronomy, but their use is hindered by the computational cost of existing inference schemes. Our scheme: requires no discretisation of the domain; scales linearly in the number of observed events; and is many orders of magnitude faster than previous sampling based approaches. The resulting algorithm is shown to outperform standard methods on synthetic examples, coal mining disaster data and in the prediction of Malaria incidences in Kenya.

Cite this Paper

BibTeX
@InProceedings{pmlr-v37-lloyd15, title = {Variational Inference for Gaussian Process Modulated Poisson Processes}, author = {Lloyd, Chris and Gunter, Tom and Osborne, Michael and Roberts, Stephen}, booktitle = {Proceedings of the 32nd International Conference on Machine Learning}, pages = {1814--1822}, year = {2015}, editor = {Bach, Francis and Blei, David}, volume = {37}, series = {Proceedings of Machine Learning Research}, address = {Lille, France}, month = {07--09 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v37/lloyd15.pdf}, url = {https://proceedings.mlr.press/v37/lloyd15.html}, abstract = {We present the first fully variational Bayesian inference scheme for continuous Gaussian-process-modulated Poisson processes. Such point processes are used in a variety of domains, including neuroscience, geo-statistics and astronomy, but their use is hindered by the computational cost of existing inference schemes. Our scheme: requires no discretisation of the domain; scales linearly in the number of observed events; and is many orders of magnitude faster than previous sampling based approaches. The resulting algorithm is shown to outperform standard methods on synthetic examples, coal mining disaster data and in the prediction of Malaria incidences in Kenya.} }
Endnote
%0 Conference Paper %T Variational Inference for Gaussian Process Modulated Poisson Processes %A Chris Lloyd %A Tom Gunter %A Michael Osborne %A Stephen Roberts %B Proceedings of the 32nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2015 %E Francis Bach %E David Blei %F pmlr-v37-lloyd15 %I PMLR %P 1814--1822 %U https://proceedings.mlr.press/v37/lloyd15.html %V 37 %X We present the first fully variational Bayesian inference scheme for continuous Gaussian-process-modulated Poisson processes. Such point processes are used in a variety of domains, including neuroscience, geo-statistics and astronomy, but their use is hindered by the computational cost of existing inference schemes. Our scheme: requires no discretisation of the domain; scales linearly in the number of observed events; and is many orders of magnitude faster than previous sampling based approaches. The resulting algorithm is shown to outperform standard methods on synthetic examples, coal mining disaster data and in the prediction of Malaria incidences in Kenya.
RIS
TY - CPAPER TI - Variational Inference for Gaussian Process Modulated Poisson Processes AU - Chris Lloyd AU - Tom Gunter AU - Michael Osborne AU - Stephen Roberts BT - Proceedings of the 32nd International Conference on Machine Learning DA - 2015/06/01 ED - Francis Bach ED - David Blei ID - pmlr-v37-lloyd15 PB - PMLR DP - Proceedings of Machine Learning Research VL - 37 SP - 1814 EP - 1822 L1 - http://proceedings.mlr.press/v37/lloyd15.pdf UR - https://proceedings.mlr.press/v37/lloyd15.html AB - We present the first fully variational Bayesian inference scheme for continuous Gaussian-process-modulated Poisson processes. Such point processes are used in a variety of domains, including neuroscience, geo-statistics and astronomy, but their use is hindered by the computational cost of existing inference schemes. Our scheme: requires no discretisation of the domain; scales linearly in the number of observed events; and is many orders of magnitude faster than previous sampling based approaches. The resulting algorithm is shown to outperform standard methods on synthetic examples, coal mining disaster data and in the prediction of Malaria incidences in Kenya. ER -
APA
Lloyd, C., Gunter, T., Osborne, M. & Roberts, S.. (2015). Variational Inference for Gaussian Process Modulated Poisson Processes. Proceedings of the 32nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 37:1814-1822 Available from https://proceedings.mlr.press/v37/lloyd15.html.

Related Material