Sequential Neural Likelihood: Fast Likelihood-free Inference with Autoregressive Flows

TL;DR

Sequential Neural Likelihood (SNL) is a novel Bayesian inference method for intractable likelihood models that uses autoregressive flows and sequential training to significantly reduce simulation costs and improve robustness and accuracy.

Contribution

SNL introduces a sequential training approach with autoregressive flows for efficient likelihood approximation in simulator-based Bayesian inference.

Findings

SNL reduces simulation costs by orders of magnitude.

SNL outperforms related neural methods in robustness and accuracy.

Provides diagnostics for calibration, convergence, and goodness-of-fit.

Abstract

We present Sequential Neural Likelihood (SNL), a new method for Bayesian inference in simulator models, where the likelihood is intractable but simulating data from the model is possible. SNL trains an autoregressive flow on simulated data in order to learn a model of the likelihood in the region of high posterior density. A sequential training procedure guides simulations and reduces simulation cost by orders of magnitude. We show that SNL is more robust, more accurate and requires less tuning than related neural-based methods, and we discuss diagnostics for assessing calibration, convergence and goodness-of-fit.