Rao-Blackwellised Particle Filtering for Dynamic Bayesian Networks

DMCA

Rao-Blackwellised Particle Filtering for Dynamic Bayesian Networks

Cached

Download Links

Other Repositories/Bibliography

DBLP

by Arnaud Doucet , Nando de Freitas , Kevin Murphy , Stuart Russell

Citations:

348 - 11 self

BibTeX

@MISC{Doucet_rao-blackwellisedparticle,
    author = {Arnaud Doucet and Nando de Freitas and Kevin Murphy and Stuart Russell},
    title = {Rao-Blackwellised Particle Filtering for Dynamic Bayesian Networks},
    year = {}
}

OpenURL

Abstract

Particle filters (PFs) are powerful sampling-based inference/learning algorithms for dynamic Bayesian networks (DBNs). They allow us to treat, in a principled way, any type of probability distribution, nonlinearity and non-stationarity. They have appeared in several fields under such names as “condensation”, “sequential Monte Carlo” and “survival of the fittest”. In this paper, we show how we can exploit the structure of the DBN to increase the efficiency of particle filtering, using a technique known as Rao-Blackwellisation. Essentially, this samples some of the variables, and marginalizes out the rest exactly, using the Kalman filter, HMM filter, junction tree algorithm, or any other finite dimensional optimal filter. We show that Rao-Blackwellised particle filters (RBPFs) lead to more accurate estimates than standard PFs. We demonstrate RBPFs on two problems, namely non-stationary online regression with radial basis function networks and robot localization and map building. We also discuss other potential application areas and provide references to some Þnite dimensional optimal filters.

Keyphrases

dynamic bayesian network rao-blackwellised particle filtering nite dimensional optimal filter potential application area finite dimensional optimal filter particle filtering accurate estimate powerful sampling-based inference learning algorithm principled way sequential monte carlo map building non-stationary online regression probability distribution rao-blackwellised particle filter hmm filter radial basis function network particle filter kalman filter junction tree algorithm several field standard pfs

DMCA