Přístupnostní navigace
E-application
Search Search Close
Publication detail
DOKOUPIL, J. VÁCLAVEK, P.
Original Title
Variable exponential forgetting for estimation of the statistics of the normal-Wishart distribution with a constant precision
Type
conference paper
Language
English
Original Abstract
The problem of estimating normal regression-type models with possibly time-varying regression parameters and constant noise precision is considered and examined from the Bayesian viewpoint. The solution we propose exploits a collaborative decision in order to face the incomplete model of parameter variations. Under this approach, a loss functional evaluating two prediction alternatives is constructed, which allows us to merge both alternatives, complying with the principles of optimization theory. Specifically, the posterior probability density function (pdf) and its flattened variant are combined by means of the geometric mean with automatically adjusted weights. The result is an automatic rescaling of the covariance matrix through the forgetting factor in response to empirically confirmed performance.
Keywords
forgetting factor; Kullback-Leibler divergence; normal-Wishart distribution
Authors
DOKOUPIL, J.; VÁCLAVEK, P.
Released
11. 12. 2019
Publisher
IEEE
Location
Nice, France
ISBN
978-1-7281-1397-5
Book
58th Conference on Decision and Control
Pages from
5094
Pages to
5100
Pages count
7
BibTex
@inproceedings{BUT160943, author="Jakub {Dokoupil} and Pavel {Václavek}", title="Variable exponential forgetting for estimation of the statistics of the normal-Wishart distribution with a constant precision", booktitle="58th Conference on Decision and Control", year="2019", pages="5094--5100", publisher="IEEE", address="Nice, France", doi="10.1109/CDC40024.2019.9029290", isbn="978-1-7281-1397-5" }