Scalable Gaussian processes for characterizing multidimensional change surfaces

William Herlands, Andrew Wilson, Hannes Nickisch, Seth Flaxman, Daniel Neill, Wilbert van Panhuis, Eric Xing

Research output: Contribution to conferencePaper

Abstract

We present a scalable Gaussian process model for identifying and characterizing smooth multidimensional changepoints, and automatically learning changes in expressive covariance structure. We use Random Kitchen Sink features to flexibly define a change surface in combination with expressive spectral mixture kernels to capture the complex statistical structure. Finally, through the use of novel methods for additive non-separable kernels, we can scale the model to large datasets. We demonstrate the model on numerical and real world data, including a large spatio-temporal disease dataset where we identify previously unknown heterogeneous changes in space and time.

Original languageEnglish (US)
Pages1013-1021
Number of pages9
StatePublished - Jan 1 2016
Event19th International Conference on Artificial Intelligence and Statistics, AISTATS 2016 - Cadiz, Spain
Duration: May 9 2016May 11 2016

Conference

Conference19th International Conference on Artificial Intelligence and Statistics, AISTATS 2016
CountrySpain
CityCadiz
Period5/9/165/11/16

Fingerprint

Gaussian Process
kernel
Nonseparable
Change Point
Covariance Structure
Gaussian Model
Large Data Sets
Process Model
Kitchens
Unknown
Model
Demonstrate
Learning

ASJC Scopus subject areas

  • Artificial Intelligence
  • Statistics and Probability

Cite this

Herlands, W., Wilson, A., Nickisch, H., Flaxman, S., Neill, D., van Panhuis, W., & Xing, E. (2016). Scalable Gaussian processes for characterizing multidimensional change surfaces. 1013-1021. Paper presented at 19th International Conference on Artificial Intelligence and Statistics, AISTATS 2016, Cadiz, Spain.

Scalable Gaussian processes for characterizing multidimensional change surfaces. / Herlands, William; Wilson, Andrew; Nickisch, Hannes; Flaxman, Seth; Neill, Daniel; van Panhuis, Wilbert; Xing, Eric.

2016. 1013-1021 Paper presented at 19th International Conference on Artificial Intelligence and Statistics, AISTATS 2016, Cadiz, Spain.

Research output: Contribution to conferencePaper

Herlands, W, Wilson, A, Nickisch, H, Flaxman, S, Neill, D, van Panhuis, W & Xing, E 2016, 'Scalable Gaussian processes for characterizing multidimensional change surfaces', Paper presented at 19th International Conference on Artificial Intelligence and Statistics, AISTATS 2016, Cadiz, Spain, 5/9/16 - 5/11/16 pp. 1013-1021.
Herlands W, Wilson A, Nickisch H, Flaxman S, Neill D, van Panhuis W et al. Scalable Gaussian processes for characterizing multidimensional change surfaces. 2016. Paper presented at 19th International Conference on Artificial Intelligence and Statistics, AISTATS 2016, Cadiz, Spain.
Herlands, William ; Wilson, Andrew ; Nickisch, Hannes ; Flaxman, Seth ; Neill, Daniel ; van Panhuis, Wilbert ; Xing, Eric. / Scalable Gaussian processes for characterizing multidimensional change surfaces. Paper presented at 19th International Conference on Artificial Intelligence and Statistics, AISTATS 2016, Cadiz, Spain.9 p.
@conference{16315e937fcf45fbaacea74ca2d583ae,
title = "Scalable Gaussian processes for characterizing multidimensional change surfaces",
abstract = "We present a scalable Gaussian process model for identifying and characterizing smooth multidimensional changepoints, and automatically learning changes in expressive covariance structure. We use Random Kitchen Sink features to flexibly define a change surface in combination with expressive spectral mixture kernels to capture the complex statistical structure. Finally, through the use of novel methods for additive non-separable kernels, we can scale the model to large datasets. We demonstrate the model on numerical and real world data, including a large spatio-temporal disease dataset where we identify previously unknown heterogeneous changes in space and time.",
author = "William Herlands and Andrew Wilson and Hannes Nickisch and Seth Flaxman and Daniel Neill and {van Panhuis}, Wilbert and Eric Xing",
year = "2016",
month = "1",
day = "1",
language = "English (US)",
pages = "1013--1021",
note = "19th International Conference on Artificial Intelligence and Statistics, AISTATS 2016 ; Conference date: 09-05-2016 Through 11-05-2016",

}

TY - CONF

T1 - Scalable Gaussian processes for characterizing multidimensional change surfaces

AU - Herlands, William

AU - Wilson, Andrew

AU - Nickisch, Hannes

AU - Flaxman, Seth

AU - Neill, Daniel

AU - van Panhuis, Wilbert

AU - Xing, Eric

PY - 2016/1/1

Y1 - 2016/1/1

N2 - We present a scalable Gaussian process model for identifying and characterizing smooth multidimensional changepoints, and automatically learning changes in expressive covariance structure. We use Random Kitchen Sink features to flexibly define a change surface in combination with expressive spectral mixture kernels to capture the complex statistical structure. Finally, through the use of novel methods for additive non-separable kernels, we can scale the model to large datasets. We demonstrate the model on numerical and real world data, including a large spatio-temporal disease dataset where we identify previously unknown heterogeneous changes in space and time.

AB - We present a scalable Gaussian process model for identifying and characterizing smooth multidimensional changepoints, and automatically learning changes in expressive covariance structure. We use Random Kitchen Sink features to flexibly define a change surface in combination with expressive spectral mixture kernels to capture the complex statistical structure. Finally, through the use of novel methods for additive non-separable kernels, we can scale the model to large datasets. We demonstrate the model on numerical and real world data, including a large spatio-temporal disease dataset where we identify previously unknown heterogeneous changes in space and time.

UR - http://www.scopus.com/inward/record.url?scp=85057321615&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85057321615&partnerID=8YFLogxK

M3 - Paper

AN - SCOPUS:85057321615

SP - 1013

EP - 1021

ER -