A STOCHASTIC ESTIMATION ALGORITHM WITH OBSERVATION AVERAGING

Authors
JUDITSKY A
Citation
A. Juditsky, A STOCHASTIC ESTIMATION ALGORITHM WITH OBSERVATION AVERAGING, IEEE transactions on automatic control, 38(5), 1993, pp. 794-798
Citations number
12
Categorie Soggetti
Controlo Theory & Cybernetics","Computer Applications & Cybernetics","Engineering, Eletrical & Electronic
Journal title
IEEE transactions on automatic control → ACNP
ISSN journal
00189286
Volume
38
Issue
5
Year of publication
1993
Pages
794 - 798
Database
ISI
SICI code
0018-9286(1993)38:5<794:ASEAWO>2.0.ZU;2-3
Abstract
In this note we present a new algorithm for the constrained problem of estimating the regression coefficients. This algorithm is based on th e idea of direct averaging of the observations in order to estimate th e search direction. It is shown that if the true parameter belongs to the ''permitted'' set then the algorithm delivers asymptotically optim al estimates of the parameter. On the other hand, finite convergence o f the method is proved when the true parameter lies outside the ''perm itted'' set.