Estimation of Change-point Models

TL;DR

This paper investigates change-point detection by analyzing Gaussian fields with trends, providing asymptotic p-value approximations for likelihood ratio tests to improve change-point estimation accuracy.

Contribution

It introduces a novel approach using Gaussian field extremes to derive asymptotic p-value approximations for change-point models.

Findings

Derived asymptotic p-value approximations for likelihood ratio statistics.

Enhanced understanding of change-point detection in Gaussian processes.

Provided theoretical foundations for more accurate change-point estimation.

Abstract

We consider the testing and estimation of change-points, locations where the distribution abruptly changes, in a sequence of observations. Motivated by this problem, in this contribution we first investigate the extremes of Gaussian fields with trend which then help us give asymptotic p-value approximations of the likelihood ratio statistics from change-point models.