Title: A greedy approach to detecting change points for functional data
Authors: Jeng-Min Chiou - Academia Sinica (Taiwan) [presenting]
Abstract: A new greedy segmentation method is presented to identify multiple change points for a functional data sequence. The estimator's consistency property holds without the common at-most-one-changepoint condition, and it is robust to the relative positions among the change points. Besides, we derive a test statistic based on the estimator and develop an algorithm to determine the number of change points. We will show the asymptotic properties of the algorithm and explore its finite sample performance through a simulation study.