In our upcoming Bayesian Working group meeting on Monday (03/03), 1:10 - 2 pm, we will have Valerie Han presenting. The talk will be virtual on zoom for which link is given below. Those on campus can join us in Snedecor 2113.

Title: Penalized Complexity Priors: Principles, Construction, and Challenges

Abstract: Priors play a crucial role in Bayesian modeling, guiding inference by incorporating domain knowledge while maintaining model flexibility. Penalized Complexity (PC) priors provide a principled framework for prior specification by ensuring shrinkage toward a simpler base model, balancing model complexity and parsimony. This talk will cover the key principles that priors should satisfy, at least according to Simpson et al 2017; the motivation behind PC priors; and the methodology for their construction. Practical challenges in computing PC priors will also be discussed, particularly in high-dimensional models.

Look forward to seeing many of you there.