Adaptive Sequential Monte Carlo for Multiple Changepoint Analysis

Heard, Nicholas A.; Turcotte, Melissa J.  M.

doi:10.1080/10618600.2016.1190281

Title: Adaptive Sequential Monte Carlo for Multiple Changepoint Analysis

Abstract

Process monitoring and control requires detection of structural changes in a data stream in real time. This paper introduces an efficient sequential Monte Carlo algorithm designed for learning unknown changepoints in continuous time. The method is intuitively simple: new changepoints for the latest window of data are proposed by conditioning only on data observed since the most recent estimated changepoint, as these observations carry most of the information about the current state of the process. The proposed method shows improved performance over the current state of the art. Another advantage of the proposed algorithm is that it can be made adaptive, varying the number of particles according to the apparent local complexity of the target changepoint probability distribution. This saves valuable computing time when changes in the changepoint distribution are negligible, and enables re-balancing of the importance weights of existing particles when a significant change in the target distribution is encountered. The plain and adaptive versions of the method are illustrated using the canonical continuous time changepoint problem of inferring the intensity of an inhomogeneous Poisson process, although the method is generally applicable to any changepoint problem. Performance is demonstrated using both conjugate and non-conjugate Bayesian models for the intensity.more »« less

Authors:

Heard, Nicholas A. ^[1]; Turcotte, Melissa J. M. ^[2]

Imperial College, London (United Kingdom). Dept. of Mathematics; Heilbronn Inst. for Mathematical Research, Bristol (United Kingdom)
Los Alamos National Lab. (LANL), Los Alamos, NM (United States)

Publication Date:: Sat May 21 00:00:00 EDT 2016

Research Org.:: Los Alamos National Laboratory (LANL), Los Alamos, NM (United States)

Sponsoring Org.:: USDOE; Engineering and Physical Sciences Research Council (EPSRC)

OSTI Identifier:: 1340927

Report Number(s):: LA-UR-16-22735
Journal ID: ISSN 1061-8600

Grant/Contract Number:: AC52-06NA25396

Resource Type:: Accepted Manuscript

Journal Name:: Journal of Computational and Graphical Statistics

Additional Journal Information:: Journal Volume: 26; Journal Issue: 2; Journal ID: ISSN 1061-8600

Publisher:: Taylor & Francis

Country of Publication:: United States

Language:: English

Subject:: 97 MATHEMATICS AND COMPUTING; Mathematics

Citation Formats


                    Heard, Nicholas A., and Turcotte, Melissa J.  M. Adaptive Sequential Monte Carlo for Multiple Changepoint Analysis.  United States: N. p., 2016. 
Web.  doi:10.1080/10618600.2016.1190281.

Copy to clipboard


                    Heard, Nicholas A., & Turcotte, Melissa J.  M. Adaptive Sequential Monte Carlo for Multiple Changepoint Analysis.  United States.  https://doi.org/10.1080/10618600.2016.1190281

Copy to clipboard


                    Heard, Nicholas A., and Turcotte, Melissa J.  M. Sat .  
"Adaptive Sequential Monte Carlo for Multiple Changepoint Analysis".  United States.  https://doi.org/10.1080/10618600.2016.1190281.  https://www.osti.gov/servlets/purl/1340927.

Copy to clipboard


                    
@article{osti_1340927,

  title        = {Adaptive Sequential Monte Carlo for Multiple Changepoint Analysis},

  author       = {Heard, Nicholas A. and Turcotte, Melissa J.  M.},

  abstractNote = {Process monitoring and control requires detection of structural changes in a data stream in real time. This paper introduces an efficient sequential Monte Carlo algorithm designed for learning unknown changepoints in continuous time. The method is intuitively simple: new changepoints for the latest window of data are proposed by conditioning only on data observed since the most recent estimated changepoint, as these observations carry most of the information about the current state of the process. The proposed method shows improved performance over the current state of the art. Another advantage of the proposed algorithm is that it can be made adaptive, varying the number of particles according to the apparent local complexity of the target changepoint probability distribution. This saves valuable computing time when changes in the changepoint distribution are negligible, and enables re-balancing of the importance weights of existing particles when a significant change in the target distribution is encountered. The plain and adaptive versions of the method are illustrated using the canonical continuous time changepoint problem of inferring the intensity of an inhomogeneous Poisson process, although the method is generally applicable to any changepoint problem. Performance is demonstrated using both conjugate and non-conjugate Bayesian models for the intensity. Lastly, appendices to the article are available online, illustrating the method on other models and applications.},

  doi          = {10.1080/10618600.2016.1190281},

  journal      = {Journal of Computational and Graphical Statistics},

  number       = 2,

  volume       = 26,

  place        = {United States},

  year         = {Sat May 21 00:00:00 EDT 2016},

  month        = {Sat May 21 00:00:00 EDT 2016}

}

Copy to clipboard

Journal Article:

Free Publicly Available Full Text

Accepted Manuscript (DOE)

Publisher's Version of Record

https://doi.org/10.1080/10618600.2016.1190281

Other availability

Search WorldCat to find libraries that may hold this journal

Citation Metrics:

Cited by: 3 works

Citation information provided by
Web of Science

Save / Share:

Export Metadata

Save to My Library

Works referenced in this record:

Convergence of Monte Carlo distribution estimates from rival samplers
journal, July 2015

Heard, Nicholas A.; Turcotte, Melissa J. M.
Statistics and Computing, Vol. 26, Issue 6
DOI: 10.1007/s11222-015-9595-0

Metropolized independent sampling with comparisons to rejection sampling and importance sampling
journal, June 1996

Liu, Jun S.
Statistics and Computing, Vol. 6, Issue 2
DOI: 10.1007/BF00162521

Reversible jump Markov chain Monte Carlo computation and Bayesian model determination
journal, January 1995

Green, Peter J.
Biometrika, Vol. 82, Issue 4
DOI: 10.1093/biomet/82.4.711

On-line inference for hidden Markov models via particle filters
journal, November 2003

Fearnhead, Paul; Clifford, Peter
Journal of the Royal Statistical Society: Series B (Statistical Methodology), Vol. 65, Issue 4
DOI: 10.1111/1467-9868.00421

Monte Carlo Filter and Smoother for Non-Gaussian Nonlinear State Space Models
journal, March 1996

Kitagawa, Genshiro
Journal of Computational and Graphical Statistics, Vol. 5, Issue 1
DOI: 10.1080/10618600.1996.10474692

On-line inference for multiple changepoint problems
journal, September 2007

Fearnhead, Paul; Liu, Zhen
Journal of the Royal Statistical Society: Series B (Statistical Methodology), Vol. 69, Issue 4
DOI: 10.1111/j.1467-9868.2007.00601.x

Adapting the Sample Size in Particle Filters Through KLD-Sampling
journal, December 2003

Fox, Dieter
The International Journal of Robotics Research, Vol. 22, Issue 12
DOI: 10.1177/0278364903022012001

Inference for a class of partially observed point process models
journal, August 2012

Martin, James S.; Jasra, Ajay; McCoy, Emma
Annals of the Institute of Statistical Mathematics, Vol. 65, Issue 3
DOI: 10.1007/s10463-012-0375-8

Following a moving target-Monte Carlo inference for dynamic Bayesian models
journal, February 2001

Gilks, Walter R.; Berzuini, Carlo
Journal of the Royal Statistical Society: Series B (Statistical Methodology), Vol. 63, Issue 1
DOI: 10.1111/1467-9868.00280

Monte Carlo Filtering of Piecewise Deterministic Processes
journal, January 2011

Whiteley, Nick; Johansen, Adam M.; Godsill, Simon
Journal of Computational and Graphical Statistics, Vol. 20, Issue 1
DOI: 10.1198/jcgs.2009.08052

Dynamic Detection of Change Points in Long Time Series
journal, June 2006

Chopin, Nicolas
Annals of the Institute of Statistical Mathematics, Vol. 59, Issue 2
DOI: 10.1007/s10463-006-0053-9

Bayesian anomaly detection methods for social networks
journal, August 2010

Heard, Nicholas A.; Weston, David J.; Platanioti, Kiriaki
The Annals of Applied Statistics, Vol. 4, Issue 2, p. 645-662
DOI: 10.1214/10-AOAS329

Sequential Monte Carlo samplers
journal, June 2006

Del Moral, Pierre; Doucet, Arnaud; Jasra, Ajay
Journal of the Royal Statistical Society: Series B (Statistical Methodology), Vol. 68, Issue 3
DOI: 10.1111/j.1467-9868.2006.00553.x

Bayesian analysis of a Poisson process with a change-point
journal, January 1986

Raftery, A. E.; Akman, V. E.
Biometrika, Vol. 73, Issue 1
DOI: 10.1093/biomet/73.1.85

Monte Carlo Filter and Smoother for Non-Gaussian Nonlinear State Space Models
journal, March 1996

Kitagawa, Genshiro
Journal of Computational and Graphical Statistics, Vol. 5, Issue 1
DOI: 10.2307/1390750

Inference for a Class of Partially Observed Point Process Models
preprint, January 2012

Martin, James S.; Jasra, Ajay; McCoy, Emma
arXiv
DOI: 10.48550/arxiv.1201.4529

Works referencing / citing this record:

Simultaneous Credible Regions for Multiple Changepoint Locations
journal, November 2018

Siems, Tobias; Hellmuth, Marc; Liebscher, Volkmar
Journal of Computational and Graphical Statistics, Vol. 28, Issue 2
DOI: 10.1080/10618600.2018.1513366

Change-Point Analysis of Asset Price Bubbles with Power-Law Hazard Function
journal, November 2019

Lynch, Christopher; Mestel, Benjamin
International Journal of Theoretical and Applied Finance, Vol. 22, Issue 07
DOI: 10.1142/s021902491950033x

Simultaneous Credible Regions for Multiple Changepoint Locations
text, January 2016

Siems, Tobias; Hellmuth, Marc; Liebscher, Volkmar
arXiv
DOI: 10.48550/arxiv.1610.04199

Stein Variational Online Changepoint Detection with Applications to Hawkes Processes and Neural Networks
preprint, January 2019

Detommaso, Gianluca; Hoitzing, Hanne; Cui, Tiangang
arXiv
DOI: 10.48550/arxiv.1901.07987

Similar Records in DOE PAGES and OSTI.GOV collections:

Anomaly Detection in Dynamic Networks

Technical Report Turcotte, Melissa

Anomaly detection in dynamic communication networks has many important security applications. These networks can be extremely large and so detecting any changes in their structure can be computationally challenging; hence, computationally fast, parallelisable methods for monitoring the network are paramount. For this reason the methods presented here use independent node and edge based models to detect locally anomalous substructures within communication networks. As a first stage, the aim is to detect changes in the data streams arising from node or edge communications. Throughout the thesis simple, conjugate Bayesian models for counting processes are used to model these data streams. Amore »« less
https://doi.org/10.2172/1160097

Full Text Available
A Surrogate-based Adaptive Sampling Approach for History Matching and Uncertainty Quantification

Conference Li, Weixuan ; Zhang, Dongxiao ; Lin, Guang

A critical procedure in reservoir simulations is history matching (or data assimilation in a broader sense), which calibrates model parameters such that the simulation results are consistent with field measurements, and hence improves the credibility of the predictions given by the simulations. Often there exist non-unique combinations of parameter values that all yield the simulation results matching the measurements. For such ill-posed history matching problems, Bayesian theorem provides a theoretical foundation to represent different solutions and to quantify the uncertainty with the posterior PDF. Lacking an analytical solution in most situations, the posterior PDF may be characterized with a samplemore »« less
https://doi.org/10.2118/173298-MS
Adaptive Learning for Concept Drift in Application Performance Modeling

Conference Madireddy, Sandeep ; Balaprakash, Prasanna ; Carns, Philip ; ...

Supervised learning is a promising approach for modeling the performance of applications running on large HPC systems. A key assumption in supervised learning is that the training and testing data are obtained under the same conditions. However, in production HPC systems these conditions might not hold because the conditions of the platform can change over time as a result of hardware degradation, hardware replacement, software upgrade, and configuration updates. These changes could alter the data distribution in a way that affects the accuracy of the predictive performance models and render them less useful; this phenomenon is referred to as conceptmore »« less
https://doi.org/10.1145/3337821.3337922

Full Text Available
Elastic functional changepoint detection of climate impacts from localized sources

Journal Article Tucker, J. Derek ; Yarger, Drew - EnvironMetrics (London)

Abstract Detecting changepoints in functional data has become an important problem as interest in monitoring of climate phenomenon has increased, where the data is functional in nature. The observed data often contains both amplitude (‐axis) and phase (‐axis) variability. If not accounted for properly, true changepoints may be undetected, and the estimated underlying mean change functions will be incorrect. In this article, an elastic functional changepoint method is developed which properly accounts for these types of variability. The method can detect amplitude and phase changepoints which current methods in the literature do not, as they focus solely on the amplitudemore »« less
https://doi.org/10.1002/env.2826
Change-point Detection and Image Segmentation for Time Series of Astrophysical Images

Journal Article Xu, Cong ; Lee, Thomas C. M. ; Günther, Hans Moritz ; ... - The Astronomical Journal (Online)

Many astrophysical phenomena are time-varying, in the sense that their intensity, energy spectrum, and/or the spatial distribution of the emission suddenly change. This paper develops a method for modeling a time series of images. Under the assumption that the arrival times of the photons follow a Poisson process, the data are binned into 4D grids of voxels (time, energy band, and x-y coordinates), and viewed as a time series of non-homogeneous Poisson images. The method assumes that at each time point, the corresponding multiband image stack is an unknown 3D piecewise constant function including Poisson noise. It also assumes thatmore »« less
https://doi.org/10.3847/1538-3881/ABE0B6

Other research related to this record: