Correlative mean-field filter for sequential and spatial data processing

Jian Gao, Tembine Hamidou

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

Abstract

Accurate and robust state estimation is a fundamental problem in signal processing. Particle filter is an effective tool to solve the filtering problem in nonlinear stochastic dynamic systems. However, when the system is mean-field dependent and the data is high-dimensional in spatial and temporal domain, the estimator may become inaccurate or even diverge. In this paper, we propose a Correlative Mean-Field (CMF) filter for a general class of nonlinear systems. The algorithm iterates in four stages: decomposition, sampling, prediction, and correction. An expectation term is incorporated into system transition model to capture the mean-field property of the sequential data. By exploring the property of the circulant matrix and its relationship with Fast Fourier Transform (FFT), sufficient virtual samples are efficiently generated by cyclic shifts of original samples in the spacial domain. The correction is modeled as an online learning problem where the sample weights are updated by the correlation output of a regression function. Optimal states are estimated by the weighted sum. We perform simulations to illustrate that under some conditions our estimator converges while traditional mean-field-free filters diverge. Finally, we implement CMF in vehicle tracking tasks and tested on 12 traffic video sequences. Experiment results show that CMF outperforms the existing mean-field-free filters.

Original languageEnglish (US)
Title of host publication17th IEEE International Conference on Smart Technologies, EUROCON 2017 - Conference Proceedings
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages243-248
Number of pages6
ISBN (Electronic)9781509038435
DOIs
StatePublished - Aug 15 2017
Event17th IEEE International Conference on Smart Technologies, EUROCON 2017 - Ohrid, Macedonia, The Former Yugoslav Republic of
Duration: Jul 6 2017Jul 8 2017

Other

Other17th IEEE International Conference on Smart Technologies, EUROCON 2017
CountryMacedonia, The Former Yugoslav Republic of
CityOhrid
Period7/6/177/8/17

Fingerprint

State estimation
Telecommunication traffic
Fast Fourier transforms
Nonlinear systems
Signal processing
Dynamical systems
Sampling
Decomposition
filters
Experiments
estimators
state estimation
Filter
nonlinear systems
learning
traffic
regression analysis
signal processing
vehicles
sampling

Keywords

  • circulant matrix
  • fourier transform
  • mean-field filter
  • vehicle tracking

ASJC Scopus subject areas

  • Computer Networks and Communications
  • Energy Engineering and Power Technology
  • Electrical and Electronic Engineering
  • Instrumentation
  • Management of Technology and Innovation
  • Artificial Intelligence
  • Information Systems
  • Signal Processing

Cite this

Gao, J., & Hamidou, T. (2017). Correlative mean-field filter for sequential and spatial data processing. In 17th IEEE International Conference on Smart Technologies, EUROCON 2017 - Conference Proceedings (pp. 243-248). [8011113] Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/EUROCON.2017.8011113

Correlative mean-field filter for sequential and spatial data processing. / Gao, Jian; Hamidou, Tembine.

17th IEEE International Conference on Smart Technologies, EUROCON 2017 - Conference Proceedings. Institute of Electrical and Electronics Engineers Inc., 2017. p. 243-248 8011113.

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

Gao, J & Hamidou, T 2017, Correlative mean-field filter for sequential and spatial data processing. in 17th IEEE International Conference on Smart Technologies, EUROCON 2017 - Conference Proceedings., 8011113, Institute of Electrical and Electronics Engineers Inc., pp. 243-248, 17th IEEE International Conference on Smart Technologies, EUROCON 2017, Ohrid, Macedonia, The Former Yugoslav Republic of, 7/6/17. https://doi.org/10.1109/EUROCON.2017.8011113
Gao J, Hamidou T. Correlative mean-field filter for sequential and spatial data processing. In 17th IEEE International Conference on Smart Technologies, EUROCON 2017 - Conference Proceedings. Institute of Electrical and Electronics Engineers Inc. 2017. p. 243-248. 8011113 https://doi.org/10.1109/EUROCON.2017.8011113
Gao, Jian ; Hamidou, Tembine. / Correlative mean-field filter for sequential and spatial data processing. 17th IEEE International Conference on Smart Technologies, EUROCON 2017 - Conference Proceedings. Institute of Electrical and Electronics Engineers Inc., 2017. pp. 243-248
@inproceedings{469c6979410d468085960441ff49c1b2,
title = "Correlative mean-field filter for sequential and spatial data processing",
abstract = "Accurate and robust state estimation is a fundamental problem in signal processing. Particle filter is an effective tool to solve the filtering problem in nonlinear stochastic dynamic systems. However, when the system is mean-field dependent and the data is high-dimensional in spatial and temporal domain, the estimator may become inaccurate or even diverge. In this paper, we propose a Correlative Mean-Field (CMF) filter for a general class of nonlinear systems. The algorithm iterates in four stages: decomposition, sampling, prediction, and correction. An expectation term is incorporated into system transition model to capture the mean-field property of the sequential data. By exploring the property of the circulant matrix and its relationship with Fast Fourier Transform (FFT), sufficient virtual samples are efficiently generated by cyclic shifts of original samples in the spacial domain. The correction is modeled as an online learning problem where the sample weights are updated by the correlation output of a regression function. Optimal states are estimated by the weighted sum. We perform simulations to illustrate that under some conditions our estimator converges while traditional mean-field-free filters diverge. Finally, we implement CMF in vehicle tracking tasks and tested on 12 traffic video sequences. Experiment results show that CMF outperforms the existing mean-field-free filters.",
keywords = "circulant matrix, fourier transform, mean-field filter, vehicle tracking",
author = "Jian Gao and Tembine Hamidou",
year = "2017",
month = "8",
day = "15",
doi = "10.1109/EUROCON.2017.8011113",
language = "English (US)",
pages = "243--248",
booktitle = "17th IEEE International Conference on Smart Technologies, EUROCON 2017 - Conference Proceedings",
publisher = "Institute of Electrical and Electronics Engineers Inc.",

}

TY - GEN

T1 - Correlative mean-field filter for sequential and spatial data processing

AU - Gao, Jian

AU - Hamidou, Tembine

PY - 2017/8/15

Y1 - 2017/8/15

N2 - Accurate and robust state estimation is a fundamental problem in signal processing. Particle filter is an effective tool to solve the filtering problem in nonlinear stochastic dynamic systems. However, when the system is mean-field dependent and the data is high-dimensional in spatial and temporal domain, the estimator may become inaccurate or even diverge. In this paper, we propose a Correlative Mean-Field (CMF) filter for a general class of nonlinear systems. The algorithm iterates in four stages: decomposition, sampling, prediction, and correction. An expectation term is incorporated into system transition model to capture the mean-field property of the sequential data. By exploring the property of the circulant matrix and its relationship with Fast Fourier Transform (FFT), sufficient virtual samples are efficiently generated by cyclic shifts of original samples in the spacial domain. The correction is modeled as an online learning problem where the sample weights are updated by the correlation output of a regression function. Optimal states are estimated by the weighted sum. We perform simulations to illustrate that under some conditions our estimator converges while traditional mean-field-free filters diverge. Finally, we implement CMF in vehicle tracking tasks and tested on 12 traffic video sequences. Experiment results show that CMF outperforms the existing mean-field-free filters.

AB - Accurate and robust state estimation is a fundamental problem in signal processing. Particle filter is an effective tool to solve the filtering problem in nonlinear stochastic dynamic systems. However, when the system is mean-field dependent and the data is high-dimensional in spatial and temporal domain, the estimator may become inaccurate or even diverge. In this paper, we propose a Correlative Mean-Field (CMF) filter for a general class of nonlinear systems. The algorithm iterates in four stages: decomposition, sampling, prediction, and correction. An expectation term is incorporated into system transition model to capture the mean-field property of the sequential data. By exploring the property of the circulant matrix and its relationship with Fast Fourier Transform (FFT), sufficient virtual samples are efficiently generated by cyclic shifts of original samples in the spacial domain. The correction is modeled as an online learning problem where the sample weights are updated by the correlation output of a regression function. Optimal states are estimated by the weighted sum. We perform simulations to illustrate that under some conditions our estimator converges while traditional mean-field-free filters diverge. Finally, we implement CMF in vehicle tracking tasks and tested on 12 traffic video sequences. Experiment results show that CMF outperforms the existing mean-field-free filters.

KW - circulant matrix

KW - fourier transform

KW - mean-field filter

KW - vehicle tracking

UR - http://www.scopus.com/inward/record.url?scp=85029380879&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85029380879&partnerID=8YFLogxK

U2 - 10.1109/EUROCON.2017.8011113

DO - 10.1109/EUROCON.2017.8011113

M3 - Conference contribution

AN - SCOPUS:85029380879

SP - 243

EP - 248

BT - 17th IEEE International Conference on Smart Technologies, EUROCON 2017 - Conference Proceedings

PB - Institute of Electrical and Electronics Engineers Inc.

ER -