Likelihood-free inference with an improved cross-entropy estimator

TL;DR

This paper introduces an improved cross-entropy estimator for likelihood-free inference that leverages augmented training data from neural network surrogate models, enhancing sample efficiency in simulation-based inference tasks.

Contribution

It extends prior work by developing a new cross-entropy estimator utilizing joint likelihood ratio and score, leading to more efficient inference with neural network surrogates.

Findings

Enhanced sample efficiency over previous methods

Effective use of joint likelihood ratio and score data

Improved inference accuracy in likelihood-free settings

Abstract

We extend recent work (Brehmer, et. al., 2018) that use neural networks as surrogate models for likelihood-free inference. As in the previous work, we exploit the fact that the joint likelihood ratio and joint score, conditioned on both observed and latent variables, can often be extracted from an implicit generative model or simulator to augment the training data for these surrogate models. We show how this augmented training data can be used to provide a new cross-entropy estimator, which provides improved sample efficiency compared to previous loss functions exploiting this augmented training data.