Role of Differential Geometry in some Bayesian Problems

Karthik Bharath (University of Nottingham)

Frank Adams 2,

A Riemannian framework is proposed to explore the utility in incorporation of differential geometric information into Bayesian models for inference. I will consider two settings to illustrate the ideas and examine the potential advantages: assessment of sensitivity of choices of prior and likelihood on posterior-based inference; Variational approximations to the posterior distribution. Results from analysis of simulated and real data will be presented.

Import this event to your Outlook calendar
▲ Up to the top