Detecting Multiple Change-Points in the Mean of Gaussian Process by Model Selection
Abstract
This paper deals with the problem of detecting the change-points in mean of a signal corrupted by an additive Gaussian noise. The number of changes and their positions are unknown. From a nonasymptotic point of view, we propose to estimate them with a method based on a penalized least-squares criterion. According to the results of Birgé and Massart, we choose the penalty function such that the resulting estimator minimizes the quadratic risk. This penalty depends on unknown constants and we propose a calibration leading to an automatic method. The performances of the method are assessed through simulation experiments. An application to real data is shown.