Online variational message passing in the hierarchical Gaussian filter

Ismail Senoz; Bert De Vries

doi:10.1109/MLSP.2018.8517019

Online variational message passing in the hierarchical Gaussian filter

Ismail Senoz, Bert De Vries

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

10 Citations (Scopus)

1 Downloads (Pure)

Abstract

We address the problem of online state and parameter estimation in hierarchical Bayesian nonlinear dynamic systems. We focus on the Hierarchical Gaussian Filter (HGF), which is a popular model in the computational neuroscience literature. For this filter, explicit equations for online state estimation (and offline parameter estimation) have been derived before. We extend this work by casting the HGF as a probabilistic factor graph and present variational message passing update rules that facilitate both online state and parameter estimation as well as online tracking of the free energy (or ELBO), which can be used as a proxy for Bayesian evidence. Due to the locality and modularity of the factor graph framework, our approach supports application of HGF's and variations as plug-in modules to a wide variety of dynamic modelling applications.

Original language	English
Title of host publication	2018 IEEE International Workshop on Machine Learning for Signal Processing, MLSP 2018 - Proceedings
Editors	Nelly Pustelnik, Zheng-Hua Tan, Zhanyu Ma, Jan Larsen
Place of Publication	Piscataway
Publisher	IEEE Computer Society
Number of pages	6
Volume	2018-September
ISBN (Electronic)	9781538654774
ISBN (Print)	978-1-5386-5478-1
DOIs	https://doi.org/10.1109/MLSP.2018.8517019
Publication status	Published - 17 Sept 2018
Event	28th IEEE International Workshop on Machine Learning for Signal Processing, MLSP 2018 - Aalborg, Denmark Duration: 17 Sept 2018 → 20 Sept 2018

Conference

Conference	28th IEEE International Workshop on Machine Learning for Signal Processing, MLSP 2018
Country/Territory	Denmark
City	Aalborg
Period	17/09/18 → 20/09/18

Keywords

Dynamical systems
Free energy
Hierarchical Gaussian Filter
Online state and parameter estimation
Variational Message Passing

Access to Document

10.1109/MLSP.2018.8517019

Cite this

@inproceedings{ce30f5a247a94b82867180a4cfe3a3ea,

title = "Online variational message passing in the hierarchical Gaussian filter",

abstract = "We address the problem of online state and parameter estimation in hierarchical Bayesian nonlinear dynamic systems. We focus on the Hierarchical Gaussian Filter (HGF), which is a popular model in the computational neuroscience literature. For this filter, explicit equations for online state estimation (and offline parameter estimation) have been derived before. We extend this work by casting the HGF as a probabilistic factor graph and present variational message passing update rules that facilitate both online state and parameter estimation as well as online tracking of the free energy (or ELBO), which can be used as a proxy for Bayesian evidence. Due to the locality and modularity of the factor graph framework, our approach supports application of HGF's and variations as plug-in modules to a wide variety of dynamic modelling applications.",

keywords = "Dynamical systems, Free energy, Hierarchical Gaussian Filter, Online state and parameter estimation, Variational Message Passing",

author = "Ismail Senoz and {De Vries}, Bert",

year = "2018",

month = sep,

day = "17",

doi = "10.1109/MLSP.2018.8517019",

language = "English",

isbn = "978-1-5386-5478-1",

volume = "2018-September",

editor = "Nelly Pustelnik and Zheng-Hua Tan and Zhanyu Ma and Jan Larsen",

booktitle = "2018 IEEE International Workshop on Machine Learning for Signal Processing, MLSP 2018 - Proceedings",

publisher = "IEEE Computer Society",

address = "United States",

note = "28th IEEE International Workshop on Machine Learning for Signal Processing, MLSP 2018 ; Conference date: 17-09-2018 Through 20-09-2018",

}

Senoz, I & De Vries, B 2018, Online variational message passing in the hierarchical Gaussian filter. in N Pustelnik, Z-H Tan, Z Ma & J Larsen (eds), 2018 IEEE International Workshop on Machine Learning for Signal Processing, MLSP 2018 - Proceedings. vol. 2018-September, 8517019, IEEE Computer Society, Piscataway, 28th IEEE International Workshop on Machine Learning for Signal Processing, MLSP 2018, Aalborg, Denmark, 17/09/18. https://doi.org/10.1109/MLSP.2018.8517019

Online variational message passing in the hierarchical Gaussian filter. / Senoz, Ismail ; De Vries, Bert.
2018 IEEE International Workshop on Machine Learning for Signal Processing, MLSP 2018 - Proceedings. ed. / Nelly Pustelnik; Zheng-Hua Tan; Zhanyu Ma; Jan Larsen. Vol. 2018-September Piscataway: IEEE Computer Society, 2018. 8517019.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

TY - GEN

T1 - Online variational message passing in the hierarchical Gaussian filter

AU - Senoz, Ismail

AU - De Vries, Bert

PY - 2018/9/17

Y1 - 2018/9/17

N2 - We address the problem of online state and parameter estimation in hierarchical Bayesian nonlinear dynamic systems. We focus on the Hierarchical Gaussian Filter (HGF), which is a popular model in the computational neuroscience literature. For this filter, explicit equations for online state estimation (and offline parameter estimation) have been derived before. We extend this work by casting the HGF as a probabilistic factor graph and present variational message passing update rules that facilitate both online state and parameter estimation as well as online tracking of the free energy (or ELBO), which can be used as a proxy for Bayesian evidence. Due to the locality and modularity of the factor graph framework, our approach supports application of HGF's and variations as plug-in modules to a wide variety of dynamic modelling applications.

AB - We address the problem of online state and parameter estimation in hierarchical Bayesian nonlinear dynamic systems. We focus on the Hierarchical Gaussian Filter (HGF), which is a popular model in the computational neuroscience literature. For this filter, explicit equations for online state estimation (and offline parameter estimation) have been derived before. We extend this work by casting the HGF as a probabilistic factor graph and present variational message passing update rules that facilitate both online state and parameter estimation as well as online tracking of the free energy (or ELBO), which can be used as a proxy for Bayesian evidence. Due to the locality and modularity of the factor graph framework, our approach supports application of HGF's and variations as plug-in modules to a wide variety of dynamic modelling applications.

KW - Dynamical systems

KW - Free energy

KW - Hierarchical Gaussian Filter

KW - Online state and parameter estimation

KW - Variational Message Passing

UR - http://www.scopus.com/inward/record.url?scp=85057068620&partnerID=8YFLogxK

U2 - 10.1109/MLSP.2018.8517019

DO - 10.1109/MLSP.2018.8517019

M3 - Conference contribution

AN - SCOPUS:85057068620

SN - 978-1-5386-5478-1

VL - 2018-September

BT - 2018 IEEE International Workshop on Machine Learning for Signal Processing, MLSP 2018 - Proceedings

A2 - Pustelnik, Nelly

A2 - Tan, Zheng-Hua

A2 - Ma, Zhanyu

A2 - Larsen, Jan

PB - IEEE Computer Society

CY - Piscataway

T2 - 28th IEEE International Workshop on Machine Learning for Signal Processing, MLSP 2018

Y2 - 17 September 2018 through 20 September 2018

ER -

Online variational message passing in the hierarchical Gaussian filter

Abstract

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this