Evaluation of variational and Markov Chain Monte Carlo methods for inference in partially observed stochastic dynamic systems

Y. Shen*, C. Archambeau, D. Cornford, M. Opper, J. Shawe-Taylor, R. Barillec

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

In recent work we have developed a novel variational inference method for partially observed systems governed by stochastic differential equations. In this paper we provide a comparison of the Variational Gaussian Process Smoother with an exact solution computed using a hybrid Monte Carlo approach to path sampling, applied to a stochastic double well potential model. It is demonstrated that the variational smoother provides us a very accurate estimate of mean path while marginal variance is slightly underestimated. We conclude with some remarks as to the advantages and disadvantages of the variational smoother.

Original languageEnglish
Title of host publicationMachine Learning for Signal Processing 17 - Proceedings of the 2007 IEEE Signal Processing Society Workshop, MLSP
PublisherIEEE
Pages306-311
Number of pages6
ISBN (Print)1424415667, 9781424415663
DOIs
Publication statusPublished - 1 Dec 2007
Event17th IEEE International Workshop on Machine Learning for Signal Processing, MLSP-2007 - Thessaloniki, United Kingdom
Duration: 27 Aug 200729 Aug 2007

Conference

Conference17th IEEE International Workshop on Machine Learning for Signal Processing, MLSP-2007
CountryUnited Kingdom
CityThessaloniki
Period27/08/0729/08/07

Fingerprint

Markov processes
Dynamical systems
Differential equations
Monte Carlo methods
Sampling

Bibliographical note

© 2007 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.

Cite this

Shen, Y., Archambeau, C., Cornford, D., Opper, M., Shawe-Taylor, J., & Barillec, R. (2007). Evaluation of variational and Markov Chain Monte Carlo methods for inference in partially observed stochastic dynamic systems. In Machine Learning for Signal Processing 17 - Proceedings of the 2007 IEEE Signal Processing Society Workshop, MLSP (pp. 306-311). [4414324] IEEE. https://doi.org/10.1109/MLSP.2007.4414324
Shen, Y. ; Archambeau, C. ; Cornford, D. ; Opper, M. ; Shawe-Taylor, J. ; Barillec, R. / Evaluation of variational and Markov Chain Monte Carlo methods for inference in partially observed stochastic dynamic systems. Machine Learning for Signal Processing 17 - Proceedings of the 2007 IEEE Signal Processing Society Workshop, MLSP. IEEE, 2007. pp. 306-311
@inproceedings{f2f8cd2fabce4eeeacadc265486181fd,
title = "Evaluation of variational and Markov Chain Monte Carlo methods for inference in partially observed stochastic dynamic systems",
abstract = "In recent work we have developed a novel variational inference method for partially observed systems governed by stochastic differential equations. In this paper we provide a comparison of the Variational Gaussian Process Smoother with an exact solution computed using a hybrid Monte Carlo approach to path sampling, applied to a stochastic double well potential model. It is demonstrated that the variational smoother provides us a very accurate estimate of mean path while marginal variance is slightly underestimated. We conclude with some remarks as to the advantages and disadvantages of the variational smoother.",
author = "Y. Shen and C. Archambeau and D. Cornford and M. Opper and J. Shawe-Taylor and R. Barillec",
note = "{\circledC} 2007 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.",
year = "2007",
month = "12",
day = "1",
doi = "10.1109/MLSP.2007.4414324",
language = "English",
isbn = "1424415667",
pages = "306--311",
booktitle = "Machine Learning for Signal Processing 17 - Proceedings of the 2007 IEEE Signal Processing Society Workshop, MLSP",
publisher = "IEEE",
address = "United States",

}

Shen, Y, Archambeau, C, Cornford, D, Opper, M, Shawe-Taylor, J & Barillec, R 2007, Evaluation of variational and Markov Chain Monte Carlo methods for inference in partially observed stochastic dynamic systems. in Machine Learning for Signal Processing 17 - Proceedings of the 2007 IEEE Signal Processing Society Workshop, MLSP., 4414324, IEEE, pp. 306-311, 17th IEEE International Workshop on Machine Learning for Signal Processing, MLSP-2007, Thessaloniki, United Kingdom, 27/08/07. https://doi.org/10.1109/MLSP.2007.4414324

Evaluation of variational and Markov Chain Monte Carlo methods for inference in partially observed stochastic dynamic systems. / Shen, Y.; Archambeau, C.; Cornford, D.; Opper, M.; Shawe-Taylor, J.; Barillec, R.

Machine Learning for Signal Processing 17 - Proceedings of the 2007 IEEE Signal Processing Society Workshop, MLSP. IEEE, 2007. p. 306-311 4414324.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Evaluation of variational and Markov Chain Monte Carlo methods for inference in partially observed stochastic dynamic systems

AU - Shen, Y.

AU - Archambeau, C.

AU - Cornford, D.

AU - Opper, M.

AU - Shawe-Taylor, J.

AU - Barillec, R.

N1 - © 2007 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.

PY - 2007/12/1

Y1 - 2007/12/1

N2 - In recent work we have developed a novel variational inference method for partially observed systems governed by stochastic differential equations. In this paper we provide a comparison of the Variational Gaussian Process Smoother with an exact solution computed using a hybrid Monte Carlo approach to path sampling, applied to a stochastic double well potential model. It is demonstrated that the variational smoother provides us a very accurate estimate of mean path while marginal variance is slightly underestimated. We conclude with some remarks as to the advantages and disadvantages of the variational smoother.

AB - In recent work we have developed a novel variational inference method for partially observed systems governed by stochastic differential equations. In this paper we provide a comparison of the Variational Gaussian Process Smoother with an exact solution computed using a hybrid Monte Carlo approach to path sampling, applied to a stochastic double well potential model. It is demonstrated that the variational smoother provides us a very accurate estimate of mean path while marginal variance is slightly underestimated. We conclude with some remarks as to the advantages and disadvantages of the variational smoother.

UR - http://www.scopus.com/inward/record.url?scp=48149106949&partnerID=8YFLogxK

UR - https://ieeexplore.ieee.org/document/4414324

U2 - 10.1109/MLSP.2007.4414324

DO - 10.1109/MLSP.2007.4414324

M3 - Conference contribution

AN - SCOPUS:48149106949

SN - 1424415667

SN - 9781424415663

SP - 306

EP - 311

BT - Machine Learning for Signal Processing 17 - Proceedings of the 2007 IEEE Signal Processing Society Workshop, MLSP

PB - IEEE

ER -

Shen Y, Archambeau C, Cornford D, Opper M, Shawe-Taylor J, Barillec R. Evaluation of variational and Markov Chain Monte Carlo methods for inference in partially observed stochastic dynamic systems. In Machine Learning for Signal Processing 17 - Proceedings of the 2007 IEEE Signal Processing Society Workshop, MLSP. IEEE. 2007. p. 306-311. 4414324 https://doi.org/10.1109/MLSP.2007.4414324