Event Details

You are kindly invited to the statistics seminar. We will host a talk by Dr. Amanda Lenzi from the University of Edinburgh.

Title: A variational neural Bayes framework for inference on intractable posterior distributions

Abstract: Recent advancements in optimization libraries and graphics processing units have enabled deep neural networks to efficiently represent massive datasets of arbitrary complexity. These improvements have motivated a new area of research in statistical inference, focusing on fast and tractable methods, particularly when traditional estimation methods fall short. In this talk, I will present a framework for Bayesian posterior estimation that learns a direct mapping from observed data to posterior distributions of model parameters, bypassing the need for explicit likelihood evaluations. The result is a fast and scalable inference method applicable to models where standard Bayesian techniques struggle. I will demonstrate the accuracy and robustness of this approach through applications such as models for spatial extreme events.