Gaussian process regression networks

Andrew Gordon Wilson; David A. Knowles; Zoubin Ghahramani

Gaussian process regression networks

Andrew Gordon Wilson, David A. Knowles, Zoubin Ghahramani

Computer Science

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

We introduce a new regression framework, Gaussian process regression networks (GPRN), which combines the structural properties of Bayesian neural networks with the nonparametric flexibility of Gaussian processes. GPRN accommodates input (predictor) dependent signal and noise correlations between multiple output (response) variables, input dependent length-scales and amplitudes, and heavy-tailed predictive distributions. We derive both elliptical slice sampling and variational Bayes inference procedures for GPRN. We apply GPRN as a multiple output regression and multivariate volatility model, demonstrating substantially improved performance over eight popular multiple output (multi-task) Gaussian process models and three multivariate volatility models on real datasets, including a 1000 dimensional gene expression dataset.

Original language	English (US)
Title of host publication	Proceedings of the 29th International Conference on Machine Learning, ICML 2012
Pages	599-606
Number of pages	8
State	Published - 2012
Event	29th International Conference on Machine Learning, ICML 2012 - Edinburgh, United Kingdom Duration: Jun 26 2012 → Jul 1 2012

Publication series

Name	Proceedings of the 29th International Conference on Machine Learning, ICML 2012
Volume	1

Other

Other	29th International Conference on Machine Learning, ICML 2012
Country/Territory	United Kingdom
City	Edinburgh
Period	6/26/12 → 7/1/12

ASJC Scopus subject areas

Human-Computer Interaction
Education

Cite this

Gaussian process regression networks. / Wilson, Andrew Gordon; Knowles, David A.; Ghahramani, Zoubin.
Proceedings of the 29th International Conference on Machine Learning, ICML 2012. 2012. p. 599-606 (Proceedings of the 29th International Conference on Machine Learning, ICML 2012; Vol. 1).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

@inproceedings{daf158c588bf44d1adb564a6c600dbfb,

title = "Gaussian process regression networks",

abstract = "We introduce a new regression framework, Gaussian process regression networks (GPRN), which combines the structural properties of Bayesian neural networks with the nonparametric flexibility of Gaussian processes. GPRN accommodates input (predictor) dependent signal and noise correlations between multiple output (response) variables, input dependent length-scales and amplitudes, and heavy-tailed predictive distributions. We derive both elliptical slice sampling and variational Bayes inference procedures for GPRN. We apply GPRN as a multiple output regression and multivariate volatility model, demonstrating substantially improved performance over eight popular multiple output (multi-task) Gaussian process models and three multivariate volatility models on real datasets, including a 1000 dimensional gene expression dataset.",

author = "Wilson, {Andrew Gordon} and Knowles, {David A.} and Zoubin Ghahramani",

year = "2012",

language = "English (US)",

isbn = "9781450312851",

series = "Proceedings of the 29th International Conference on Machine Learning, ICML 2012",

pages = "599--606",

booktitle = "Proceedings of the 29th International Conference on Machine Learning, ICML 2012",

note = "29th International Conference on Machine Learning, ICML 2012 ; Conference date: 26-06-2012 Through 01-07-2012",

}

TY - GEN

T1 - Gaussian process regression networks

AU - Wilson, Andrew Gordon

AU - Knowles, David A.

AU - Ghahramani, Zoubin

PY - 2012

Y1 - 2012

N2 - We introduce a new regression framework, Gaussian process regression networks (GPRN), which combines the structural properties of Bayesian neural networks with the nonparametric flexibility of Gaussian processes. GPRN accommodates input (predictor) dependent signal and noise correlations between multiple output (response) variables, input dependent length-scales and amplitudes, and heavy-tailed predictive distributions. We derive both elliptical slice sampling and variational Bayes inference procedures for GPRN. We apply GPRN as a multiple output regression and multivariate volatility model, demonstrating substantially improved performance over eight popular multiple output (multi-task) Gaussian process models and three multivariate volatility models on real datasets, including a 1000 dimensional gene expression dataset.

AB - We introduce a new regression framework, Gaussian process regression networks (GPRN), which combines the structural properties of Bayesian neural networks with the nonparametric flexibility of Gaussian processes. GPRN accommodates input (predictor) dependent signal and noise correlations between multiple output (response) variables, input dependent length-scales and amplitudes, and heavy-tailed predictive distributions. We derive both elliptical slice sampling and variational Bayes inference procedures for GPRN. We apply GPRN as a multiple output regression and multivariate volatility model, demonstrating substantially improved performance over eight popular multiple output (multi-task) Gaussian process models and three multivariate volatility models on real datasets, including a 1000 dimensional gene expression dataset.

UR - http://www.scopus.com/inward/record.url?scp=84867133704&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84867133704&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:84867133704

SN - 9781450312851

T3 - Proceedings of the 29th International Conference on Machine Learning, ICML 2012

SP - 599

EP - 606

BT - Proceedings of the 29th International Conference on Machine Learning, ICML 2012

T2 - 29th International Conference on Machine Learning, ICML 2012

Y2 - 26 June 2012 through 1 July 2012

ER -

Gaussian process regression networks

Abstract

Publication series

Other

ASJC Scopus subject areas

Other files and links

Fingerprint

Cite this