Almost optimal sequential tests of discrete composite hypotheses

Georgios Fellouris, Alexander G. Tartakovsky

Research output: Contribution to journalArticlepeer-review

Abstract

We consider the problem of sequentially testing a simple null hypothesis, H0, versus a composite alternative hypothesis, H1, that consists of a finite set of densities. We study sequential tests that are based on thresholding of mixturebased likelihood ratio statistics and weighted generalized likelihood ratio statistics. It is shown that both sequential tests have several asymptotic optimality properties as error probabilities go to zero. First, for any weights, they minimize the expected sample size within a constant term under every scenario in H1 and at least to first order under H1. Second, for appropriate weights that are specified up to a prior distribution, they minimize a weighted expected sample size in H 1 within an asymptotically negligible term. Third, for a particular prior distribution, they are almost minimax with respect to the expected Kullback-Leibler divergence until stopping. Furthermore, based on high-order asymptotic expansions for the operating characteristics, we propose prior distributions that lead to a robust behavior. Finally, based on asymptotic analysis as well as on simulation experiments, we argue that both tests have the same performance when they are designed with the same weights.

Original languageEnglish (US)
Pages (from-to)1717-1741
Number of pages25
JournalStatistica Sinica
Volume23
Issue number4
DOIs
StatePublished - Jul 1 2013
Externally publishedYes

Keywords

  • Asymptotic optimality
  • Generalized likelihood ratio
  • Minimax sequential tests
  • Mixture-based tests

ASJC Scopus subject areas

  • Statistics and Probability
  • Statistics, Probability and Uncertainty

Fingerprint

Dive into the research topics of 'Almost optimal sequential tests of discrete composite hypotheses'. Together they form a unique fingerprint.

Cite this