Title: Variance estimation for change-point model
Authors: Yue Niu - University of Arizona (United States) [presenting]
Abstract: The variance of noise plays an important role in many change-point detection methods. For example, in binary segmentation and related methods, the variance is required to decide when to stop the procedure. In practice, people usually use conventional methods to estimate the variance based on residuals. However, these methods may be problematic when there are many change points. We will introduce a variance estimation method and show its advantage.