Abstract : This dissertation focuses on sequential decision procedures to detect changes in the statistical model of an observed random process when these changes can occur at unknown times. In the disorder problem, the samples are drawn according to one statistical model until some unknown time after which all the samples correspond to a second model. In the transient problem, the model then reverts back to the first one after some finite time. A detection procedure known as Page's test is investigated for the quick detection of the disorder. A simple asymptotic measure is defined and an analytic formula is developed which can be useful in evaluating the performance of Page's test in various situations. By examining the local performance, it is found that the performance is directly related to the efficacy in the binary hypothesis testing situation, allowing the wealth of results in that context to be transferred to Page's test. Because Page's test appears to detect changes in distribution quickly, it is a candidate for transient signal detection.
Stored only when both papers are in the corpus — 35 citations exist upstream.
Imported from OpenAlex (CC0) — nothing here is generated.