Oracle Wiener filtering of a Gaussian signal

A. Babenko, E. Belitser

Research output: Contribution to journalArticleAcademicpeer-review

Abstract

We study the problem of filtering a Gaussian process whose trajectories, in some sense, have an unknown smoothness ß0 from the white noise of small intensity e. If we knew the parameter ß0, we would use the Wiener filter which has the meaning of oracle. Our goal is now to mimic the oracle, i.e., construct such a filter without the knowledge of the smoothness parameter ß0 that has the same quality (at least with respect to the convergence rate) as the oracle. It is known that in the pointwise minimax estimation, the adaptive minimax rate is worse by a log factor as compared to the nonadaptive one. By constructing a filter which mimics the oracle Wiener filter, we show that there is no loss of quality in terms of rate for the Bayesian counterpart of this problem – adaptive filtering problem.
Original languageEnglish
Pages (from-to)16-24
JournalTheory of Stochastic Processes
Volume17(33)
Issue number2
Publication statusPublished - 2011

Fingerprint

Dive into the research topics of 'Oracle Wiener filtering of a Gaussian signal'. Together they form a unique fingerprint.

Cite this