Recursive estimation: asymptotic confidence regions by empirical quantiles
作者:
Jürgen Dippon,
期刊:
Sequential Analysis
(Taylor Available online 1992)
卷期:
Volume 11,
issue 1
页码: 91-100
ISSN:0747-4946
年代: 1992
DOI:10.1080/07474949208836246
出版商: Marcel Dekker, Inc.
关键词: stochastic approximation;Banach space;limit distribution;Wiener-Hopf equation;stopping time
数据来源: Taylor
摘要:
Central limit theorems in stochastic approximation theory often involve an integral of a limit process of the observation errors. This integral is approximated by a further stochastic iteration. Then empirical quantiles are used to construct spherical asymptotic confidence regions for the parameter to be estimated. As an example the method is applied to a recursive solution of a linear equation in a separable real Banach space.
点击下载:
PDF (263KB)
返 回