The Bernstein-von Mises Theorem for Semiparametric Mixture

Explore the Bernstein-von Mises theorem for semiparametric mixtures in this 57-minute seminar presented by Dr. Stefan Franssen from CNRS, Université Sorbonne Paris Nord. Delve into advanced statistical theory as part of the "Representing, calibrating & leveraging prediction uncertainty from statistics to machine learning" event at the Isaac Newton Institute. Examine the theoretical foundations and applications of this important theorem in the context of semiparametric mixture models, gaining insights into how classical asymptotic results extend to more complex statistical frameworks. Learn about the intersection of Bayesian statistics and semiparametric methods, with particular focus on posterior concentration and asymptotic normality properties. Discover how these theoretical developments contribute to understanding prediction uncertainty in modern statistical and machine learning applications.