Adaptive approximate filtering of state-space models

Result code in IS VaVaI
<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F67985556%3A_____%2F15%3A00447270" target="_blank" >RIV/67985556:_____/15:00447270 - isvavai.cz</a>
Result on the web
—
DOI - Digital Object Identifier
—

Result language
angličtina
Original language name
Adaptive approximate filtering of state-space models
Original language description
Approximate Bayesian computation (ABC) filtration of state-space models replaces popular particle filters in cases where the observation models (i.e. likelihoods) are either computationally too demanding or completely intractable, but it is still possible to simulate from them. These sequential Monte Carlo methods evaluate importance weights based on the distance between the true observation and the simulated pseudo-observations. The paper proposes a new adaptive method consisting of probability kernel-based evaluation of importance weights with online determination of kernel scale. It is shown that the resulting algorithm achieves performance close to particle filters in the case of well-specified models, and outperforms generic particle filters and state-of-art ABC filters under heavy-tailed noise and model misspecification.
Czech name
—
Czech description
—

Project
<a href="/en/project/GP14-06678P" target="_blank" >GP14-06678P: Distributed dynamic estimation in diffusion networks</a><br>
Continuities
I - Institucionalni podpora na dlouhodoby koncepcni rozvoj vyzkumne organizace

Publication year
2015
Confidentiality
S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Article name in the collection
Proceedings of 23rd European Signal Processing Conference
ISBN
978-0-9928626-4-0
ISSN
2076-1465
e-ISSN
—
Number of pages
5
Pages from-to
2236-2240
Publisher name
EURASIP
Place of publication
Nice
Event location
Nice
Event date
Aug 31, 2015
Type of event by nationality
WRD - Celosvětová akce
UT code for WoS article
—

Similar results(10)