Likelihood-free inference with an improved cross-entropy estimator

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.

arXiv comment8 pages, 3 figures

References in corpus (15)

• WaveNet: A Generative Model for Raw Audio
• Conditional Image Generation with PixelCNN Decoders
• Pixel Recurrent Neural Networks
• NICE: Non-linear Independent Components Estimation
• Variational Inference with Normalizing Flows
• Density estimation using Real NVP
• Mining gold from implicit models to improve likelihood-free inference
• Constraining Effective Field Theories with Machine Learning
• Massive optimal data compression and density estimation for scalable, likelihood-free inference in cosmology
• A Guide to Constraining Effective Field Theories with Machine Learning
• Fast $\epsilon$-free Inference of Simulation Models with Bayesian Conditional Density Estimation
• Approximating Likelihood Ratios with Calibrated Discriminative Classifiers
• Hierarchical Implicit Models and Likelihood-Free Variational Inference
• Inference Networks for Sequential Monte Carlo in Graphical Models
• Adversarial Variational Optimization of Non-Differentiable Simulators

Cited by in corpus (9)

• Sequential Neural Likelihood: Fast Likelihood-free Inference with Autoregressive Flows
• Parametrized classifiers for optimal EFT sensitivity
• Minimax Confidence Intervals for the Sliced Wasserstein Distance
• Effective LHC measurements with matrix elements and machine learning
• Matrix Element Regression with Deep Neural Networks -- breaking the CPU barrier
• Resonance Searches with Machine Learned Likelihood Ratios
• Parameter Inference from Event Ensembles and the Top-Quark Mass
• Neural Density Estimation and Likelihood-free Inference
• Dealing with Nuisance Parameters using Machine Learning in High Energy Physics: a Review