Mathematics Colloquium
Jürgen Jost
Max Planck Institute
Title: The Geometry of Parametric and Bayesian Statistics
Date: Friday, October 15, 2021
Place and Time: Zoom, 3:05-3:55 pm
Abstract.
For a general and abstract mathematical treatment of parametric or Bayesian statistics, one needs some structure on the space of probability measures on a sample space. For parametric statistics, one needs a metric that has natural invariance properties. This is the Fisher-Rao metric which is studied in information geometry. For Bayesian statistics, one looks at measures on the space of probability measures. In this talk, I shall explain these concepts and present recent insight.