Student's t-Filters for Noise Scale Estimation

Filip Tronarp; Toni Karvonen; Simo Särkkä

doi:10.1109/LSP.2018.2889440

Student's t-Filters for Noise Scale Estimation

Filip Tronarp, Toni Karvonen, Simo Särkkä

Tutkimustuotos: Lehtiartikkeli › Article › Scientific › vertaisarvioitu

16 Sitaatiot (Scopus)

337 Lataukset (Pure)

Abstrakti

In this letter, we analyze certain student's t-filters for linear Gaussian systems with misspecified noise covariances. It is shown that under appropriate conditions, the filter both estimates the state and re-scales the noise covariance matrices in a Kullback-Leibler optimal fashion. If the noise covariances are misscaled by a common scalar, then the re-scaling is asymptotically exact. We also compare the student's t.-filter scale estimates to the maximum-likelihood estimates. Simulations demonstrating the results on the Wiener velocity model are provided.

Alkuperäiskieli	Englanti
Artikkeli	8606947
Sivut	352-356
Sivumäärä	5
Julkaisu	IEEE Signal Processing Letters
Vuosikerta	26
Numero	2
DOI - pysyväislinkit	https://doi.org/10.1109/LSP.2018.2889440
Tila	Julkaistu - 1 helmik. 2019
OKM-julkaisutyyppi	A1 Alkuperäisartikkeli tieteellisessä aikakauslehdessä

Pääsy asiakirjaan

10.1109/LSP.2018.2889440

ELEC_Tronarp_etal_Students-t-Filters-for-Noise_IEEESigProLet_26_2_2019Accepted author manuscript, 369 KB

Muut tiedostot ja linkit

Link to publication in Scopus

Probabilistinen syväoppiminen hierarkkisilla stokastisilla osittaisdifferentiaaliyhtälöillä
Särkkä, S., Karvonen, T., Sarmavuori, J., Raitoharju, M., Bahrami Rad, A., Hostettler, R., Emzir, M., Gao, R., Purisha, Z. & Tronarp, F.
01/01/2018 → 31/12/2019
Projekti: Academy of Finland: Other research funding
Sekventiaalisia Monte Carlo -menetelmiä tila- ja parametriestimointiin stokastisissa dynaamisissa systeemeissä
Särkkä, S.
01/06/2015 → 31/08/2018
Projekti: Academy of Finland: Other research funding

Siteeraa tätä

@article{9c85310f4c5d4b3ba5472a256b6305d9,

title = "Student's t-Filters for Noise Scale Estimation",

abstract = "In this letter, we analyze certain student's t-filters for linear Gaussian systems with misspecified noise covariances. It is shown that under appropriate conditions, the filter both estimates the state and re-scales the noise covariance matrices in a Kullback-Leibler optimal fashion. If the noise covariances are misscaled by a common scalar, then the re-scaling is asymptotically exact. We also compare the student's t.-filter scale estimates to the maximum-likelihood estimates. Simulations demonstrating the results on the Wiener velocity model are provided.",

keywords = "Kalman, Kalman filtering, Variances, model mis-specification, noise covariance estimation, student's t-filtering",

author = "Filip Tronarp and Toni Karvonen and Simo S{\"a}rkk{\"a}",

year = "2019",

month = feb,

day = "1",

doi = "10.1109/LSP.2018.2889440",

language = "English",

volume = "26",

pages = "352--356",

journal = "IEEE Signal Processing Letters",

issn = "1070-9908",

publisher = "IEEE",

number = "2",

}

TY - JOUR

T1 - Student's t-Filters for Noise Scale Estimation

AU - Tronarp, Filip

AU - Karvonen, Toni

AU - Särkkä, Simo

PY - 2019/2/1

Y1 - 2019/2/1

N2 - In this letter, we analyze certain student's t-filters for linear Gaussian systems with misspecified noise covariances. It is shown that under appropriate conditions, the filter both estimates the state and re-scales the noise covariance matrices in a Kullback-Leibler optimal fashion. If the noise covariances are misscaled by a common scalar, then the re-scaling is asymptotically exact. We also compare the student's t.-filter scale estimates to the maximum-likelihood estimates. Simulations demonstrating the results on the Wiener velocity model are provided.

AB - In this letter, we analyze certain student's t-filters for linear Gaussian systems with misspecified noise covariances. It is shown that under appropriate conditions, the filter both estimates the state and re-scales the noise covariance matrices in a Kullback-Leibler optimal fashion. If the noise covariances are misscaled by a common scalar, then the re-scaling is asymptotically exact. We also compare the student's t.-filter scale estimates to the maximum-likelihood estimates. Simulations demonstrating the results on the Wiener velocity model are provided.

KW - Kalman

KW - Kalman filtering

KW - Variances

KW - model mis-specification

KW - noise covariance estimation

KW - student's t-filtering

UR - http://www.scopus.com/inward/record.url?scp=85060530856&partnerID=8YFLogxK

U2 - 10.1109/LSP.2018.2889440

DO - 10.1109/LSP.2018.2889440

M3 - Article

SN - 1070-9908

VL - 26

SP - 352

EP - 356

JO - IEEE Signal Processing Letters

JF - IEEE Signal Processing Letters

IS - 2

M1 - 8606947

ER -

Student's t-Filters for Noise Scale Estimation

Abstrakti

Pääsy asiakirjaan

Muut tiedostot ja linkit

Sormenjälki

Projektit

Probabilistinen syväoppiminen hierarkkisilla stokastisilla osittaisdifferentiaaliyhtälöillä

Sekventiaalisia Monte Carlo -menetelmiä tila- ja parametriestimointiin stokastisissa dynaamisissa systeemeissä

Siteeraa tätä