Recursive least-squares and accelerated convergence in stochastic approximation schemes (Q2722623)
From MaRDI portal
 | This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use this page instead for the normal view: Recursive least-squares and accelerated convergence in stochastic approximation schemes |
scientific article; zbMATH DE number 1613242
| Language | Label | Description | Also known as |
|---|---|---|---|
| English | Recursive least-squares and accelerated convergence in stochastic approximation schemes |
scientific article; zbMATH DE number 1613242 |
Statements
Recursive least-squares and accelerated convergence in stochastic approximation schemes (English)
0 references
2 July 2001
0 references
linear regression model
0 references
time-varying parameters
0 references
stochastic approximation
0 references
recursive least squares
0 references
0 references
The paper deals with a (generalized) linear regression model in which a parameter is varying in time, however a case of white regressors is considered only. The aim of the paper is to deal with an acceleration of the convergence of approximation (gradient based) algorithms.NEWLINENEWLINEThe introduced accelerated scheme is obtained from the basic algorithm by a second round averaging. A comparison with the recursive-least-squares estimate is introduced in the paper. Especially, the corresponding asymptotic result is formulated in a Theorem that is carefully proved.NEWLINENEWLINEThe paper follows a few former papers on the similar topic. Moreover, in spite of the fact that only white regression is admitted in suggested methods, a discussion about a possibility of mixing regressors can be there found.
0 references
