Estimating Marginal Likelihoods in Likelihood-Free Inference via Neural Density Estimation
By: Paul Bastide , Arnaud Estoup , Jean-Michel Marin and more
Potential Business Impact:
Estimates how well a computer model fits data.
The marginal likelihood, or evidence, plays a central role in Bayesian model selection, yet remains notoriously challenging to compute in likelihood-free settings. While Simulation-Based Inference (SBI) techniques such as Sequential Neural Likelihood Estimation (SNLE) offer powerful tools to approximate posteriors using neural density estimators, they typically do not provide estimates of the evidence. In this technical report presented at BayesComp 2025, we present a simple and general methodology to estimate the marginal likelihood using the output of SNLE.
Similar Papers
Reducing Calls to the Simulator in Simulation Based Inference (SBI)
Computation
Makes computer models learn faster with less data.
Neural Methods for Multiple Systems Estimation Models
Applications
Helps count hidden people faster and better.
Multilevel neural simulation-based inference
Machine Learning (Stat)
Makes computer models learn faster with cheaper versions.