On robust stopping times for detecting changes in distribution
From MaRDI portal
Publication:6300740
arXiv1804.09014MaRDI QIDQ6300740FDOQ6300740
Authors: Yu. Golubev, Mher M. Safarian
Publication date: 24 April 2018
Abstract: Let be independent random variables observed sequentially and such that have a common probability density , while are all distributed according to . It is assumed that and are known, but the time change is unknown and the goal is to construct a stopping time that detects the change-point as soon as possible. The existing approaches to this problem rely essentially on some a priori information about . For instance, in Bayes approaches, it is assumed that is a random variable with a known probability distribution. In methods related to hypothesis testing, this a priori information is hidden in the so-called average run length. The main goal in this paper is to construct stopping times which do not make use of a priori information about , but have nearly Bayesian detection delays. More precisely, we propose stopping times solving approximately the following problem: �egin{equation*} �egin{split} &quad Delta( heta; au^alpha)
ightarrowmin_{ au^alpha}quad extbf{subject to}quad alpha( heta; au^alpha)le alpha extbf{ for any} hetage1, end{split} end{equation*} where is extit{the false alarm probability} and is extit{the average detection delay}, %In this paper, we construct such that %[ % max_{ hetage 1}alpha( heta;widetilde{ au}^alpha)le alpha ext{and} %Delta( heta;widetilde{ au}^alpha)le (1+o(1))log( heta/alpha), ext{as} heta/alpha%
ightarrowinfty, %] and explain why such stopping times are robust w.r.t. a priori information about .
This page was built for publication: On robust stopping times for detecting changes in distribution
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6300740)