Variance reduction for particle filters of systems with time scale separation

Dror Givon, Panagiotis Stinis, Jonathan Weare

Research output: Contribution to journalArticle

Abstract

We present a particle filter construction for a system that exhibits time-scale separation. The separation of time scales allows two simplifications that we exploit: 1) the use of the averaging principle for the dimensional reduction of the dynamics for each particle during the prediction step and 2) the factorization of the transition probability for the Rao-Blackwellization of the update step. The resulting particle filter is faster and has smaller variance than the particle filter based on the original system. The method is tested on a multiscale stochastic differential equation and on a multiscale pure jump diffusion motivated by chemical reactions.

Original languageEnglish (US)
Pages (from-to)424-435
Number of pages12
JournalIEEE Transactions on Signal Processing
Volume57
Issue number2
DOIs
StatePublished - Feb 25 2009

Fingerprint

Factorization
Chemical reactions
Differential equations

Keywords

  • Dimensional reduction
  • Jump Markov processes
  • Multiscale
  • Particle filter
  • Rao - Blackwellization
  • Stochastic differential equations
  • Variance reduction

ASJC Scopus subject areas

  • Signal Processing
  • Electrical and Electronic Engineering

Cite this

Variance reduction for particle filters of systems with time scale separation. / Givon, Dror; Stinis, Panagiotis; Weare, Jonathan.

In: IEEE Transactions on Signal Processing, Vol. 57, No. 2, 25.02.2009, p. 424-435.

Research output: Contribution to journalArticle

Givon, Dror ; Stinis, Panagiotis ; Weare, Jonathan. / Variance reduction for particle filters of systems with time scale separation. In: IEEE Transactions on Signal Processing. 2009 ; Vol. 57, No. 2. pp. 424-435.
@article{1add8baaadf541879f9fc7958d0fd8d1,
title = "Variance reduction for particle filters of systems with time scale separation",
abstract = "We present a particle filter construction for a system that exhibits time-scale separation. The separation of time scales allows two simplifications that we exploit: 1) the use of the averaging principle for the dimensional reduction of the dynamics for each particle during the prediction step and 2) the factorization of the transition probability for the Rao-Blackwellization of the update step. The resulting particle filter is faster and has smaller variance than the particle filter based on the original system. The method is tested on a multiscale stochastic differential equation and on a multiscale pure jump diffusion motivated by chemical reactions.",
keywords = "Dimensional reduction, Jump Markov processes, Multiscale, Particle filter, Rao - Blackwellization, Stochastic differential equations, Variance reduction",
author = "Dror Givon and Panagiotis Stinis and Jonathan Weare",
year = "2009",
month = "2",
day = "25",
doi = "10.1109/TSP.2008.2008252",
language = "English (US)",
volume = "57",
pages = "424--435",
journal = "IEEE Transactions on Signal Processing",
issn = "1053-587X",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "2",

}

TY - JOUR

T1 - Variance reduction for particle filters of systems with time scale separation

AU - Givon, Dror

AU - Stinis, Panagiotis

AU - Weare, Jonathan

PY - 2009/2/25

Y1 - 2009/2/25

N2 - We present a particle filter construction for a system that exhibits time-scale separation. The separation of time scales allows two simplifications that we exploit: 1) the use of the averaging principle for the dimensional reduction of the dynamics for each particle during the prediction step and 2) the factorization of the transition probability for the Rao-Blackwellization of the update step. The resulting particle filter is faster and has smaller variance than the particle filter based on the original system. The method is tested on a multiscale stochastic differential equation and on a multiscale pure jump diffusion motivated by chemical reactions.

AB - We present a particle filter construction for a system that exhibits time-scale separation. The separation of time scales allows two simplifications that we exploit: 1) the use of the averaging principle for the dimensional reduction of the dynamics for each particle during the prediction step and 2) the factorization of the transition probability for the Rao-Blackwellization of the update step. The resulting particle filter is faster and has smaller variance than the particle filter based on the original system. The method is tested on a multiscale stochastic differential equation and on a multiscale pure jump diffusion motivated by chemical reactions.

KW - Dimensional reduction

KW - Jump Markov processes

KW - Multiscale

KW - Particle filter

KW - Rao - Blackwellization

KW - Stochastic differential equations

KW - Variance reduction

UR - http://www.scopus.com/inward/record.url?scp=60549104081&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=60549104081&partnerID=8YFLogxK

U2 - 10.1109/TSP.2008.2008252

DO - 10.1109/TSP.2008.2008252

M3 - Article

AN - SCOPUS:60549104081

VL - 57

SP - 424

EP - 435

JO - IEEE Transactions on Signal Processing

JF - IEEE Transactions on Signal Processing

SN - 1053-587X

IS - 2

ER -