Fast, accurate and lightweight sequential simulation-based inference using Gaussian locally linear mappings

TL;DR

This paper introduces a lightweight, structured mixture-based approach for simulation-based Bayesian inference that achieves high accuracy and efficiency, outperforming neural network methods especially on complex, multimodal posteriors.

Contribution

The authors develop a novel structured mixture model for likelihood and posterior approximation, offering a more computationally efficient alternative to neural network-based SBI methods.

Findings

Achieves comparable or better accuracy than neural network methods.

Significantly reduces computational cost.

Effectively handles multimodal posteriors.

Abstract

Bayesian inference for complex models with an intractable likelihood can be tackled using algorithms performing many calls to computer simulators. These approaches are collectively known as "simulation-based inference" (SBI). Recent SBI methods have made use of neural networks (NN) to provide approximate, yet expressive constructs for the unavailable likelihood function and the posterior distribution. However, the trade-off between accuracy and computational demand leaves much space for improvement. In this work, we propose an alternative that provides both approximations to the likelihood and the posterior distribution, using structured mixtures of probability distributions. Our approach produces accurate posterior inference when compared to state-of-the-art NN-based SBI methods, even for multimodal posteriors, while exhibiting a much smaller computational footprint. We illustrate our results on several benchmark models from the SBI literature and on a biological model of the translation kinetics after mRNA transfection.