The Hidden Geometry of Hierarchical Models

About this episode

Alex and Stefan discuss the complexities of hierarchical models in Bayesian statistics and their implications for simulation-based inference.

Today's clip is from Episode 157 featuring Stefan Radev. In this conversation, Alex and Stefan dig into one of the hardest open problems in simulation-based inference — hierarchical models. The core idea: when you move from flat to hierarchical models, you're no longer estimating one set of parameters. You have local parameters that vary by location (or subject, or city) and global parameters that capture what's shared across all of them. And you don't just want each separately — you want the full joint posterior, because that's where the Bayesian magic of shrinkage actually lives. Stefan builds the problem from the ground up. Start with the simplest hierarchical case: a two-level model. He uses electoral forecasting in France as the example — cities nested inside departments nested inside the whole country. Now your simulator has to cover all three levels. If that simulator is slow (think: brain emulators, minutes per sample), scaling to hundreds of groups becomes completely intractable. Memory issues, specialized network requirements, the works. The key insight: this problem has structure you can exploit. The joint posterior factorizes in a particularly nice way — each local…

People in this episode

Host: Alexandre Andorra

Guest: Stefan Radev

Topics covered

• hierarchical models
• Bayesian statistics
• simulation-based inference
• electoral forecasting
• joint posterior
• local parameters
• global parameters

Keywords

• hierarchical models
• Bayesian magic
• joint posterior
• electoral forecasting
• local parameters
• global parameters
• simulation

Mentioned in this episode

Places: France

More episodes of Learning Bayesian Statistics

• Exact GPs vs Approximations: When to Use Each (and Why It Matters) · June 10, 2026 · 4 min
• #159 Bayesian Occupancy Models, with Matthijs Hollanders · June 8, 2026 · 1h 26m
• Can AI Learn What Experts Know? Automating Prior Elicitation with Generative Models · June 2, 2026 · 5 min
• #158 Bayesian Workflows & Foundation Models, with Stefan Radev · May 21, 2026 · 1h 19m
• #157 Amortized Inference & BayesFlow in Practice, with Stefan Radev · May 6, 2026 · 1h 19m
• How to Design Better Experiments with Expected Information Gain · May 1, 2026 · 6 min