Measure-transformed Gaussian quasi score test in the presence of nuisance parameters

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

In this paper, we extend the measure-transformed Gaussian quasi score test (MT-GQST) for the case where nuisance parameters are present. The proposed extension is based on a zero-expectation property of a partial Gaussian quasi score function under the transformed null distribution. The nuisance parameters are estimated under the null hypothesis via the measure-transformed Gaussian quasi MLE. In the paper, we analyze the effect of the probability measure-transformation on the asymptotic detection performance of the extended MT-GQST. This leads to a data-driven procedure for selection of the generating function of the considered transform, called MT-function, which, in practice, weights the data points. Furthermore, we provide conditions on the MT-function to ensure stability of the asymptotic false-alarm-rate in the presence of noisy outliers. The extended MT-GQST is applied for testing a vector parameter of interest comprising a noisy multivariate linear data model in the presence of nuisance parameters. Simulation study illustrates its advantages over other robust detectors.

Original languageAmerican English
Title of host publicationEUSIPCO 2019 - 27th European Signal Processing Conference
ISBN (Electronic)9789082797039
DOIs
StatePublished - 1 Sep 2019
Event27th European Signal Processing Conference, EUSIPCO 2019 - A Coruna, Spain
Duration: 2 Sep 20196 Sep 2019

Publication series

NameEuropean Signal Processing Conference
Volume2019-September

Conference

Conference27th European Signal Processing Conference, EUSIPCO 2019
Country/TerritorySpain
CityA Coruna
Period2/09/196/09/19

All Science Journal Classification (ASJC) codes

  • Signal Processing
  • Electrical and Electronic Engineering

Fingerprint

Dive into the research topics of 'Measure-transformed Gaussian quasi score test in the presence of nuisance parameters'. Together they form a unique fingerprint.

Cite this