Event Details

You are kindly invited to the statistics seminar. We will host a talk by Dr Lorenzo Cappello from Universitat Pompeu Fabra.

Title: A discussion on the use of Bayesian inference for change point detection

Abstract: Change point detection has garnered significant interest over the past decade, yet most research has focused on non-Bayesian approaches. This trend is driven by several factors, including the computational demands of Bayesian methods and the challenges of establishing theoretical guarantees for their performance. In this talk, I will discuss two scenarios where Bayesian methods offer distinct advantages in change point detection. First, when quantifying uncertainty in the change point location is a primary goal. Second, when the model within-segment assumes a complex parametric form. I will present results demonstrating the benefits of Bayesian approaches in these settings and discuss ideas to address the computational and theoretical challenges associated with their use. Based on joint works with Oscar Hernan Madrid Padilla (UCLA), Davis Berlind (UCLA), and Nicolas Bianco (Karlsruhe).