Dynamic Likelihood-free Inference via Ratio Estimation (DIRE)

Abstract

Parametric statistical models that are implicitly defined in terms of a stochastic data generating process are used in a wide range of scientific disciplines because they enable accurate modeling. However, learning the parameters from observed data is generally very difficult because their likelihood function is typically intractable. Likelihood-free Bayesian inference methods have been proposed which include the frameworks of approximate Bayesian computation (ABC), synthetic likelihood, and its recent generalization that performs likelihood-free inference by ratio estimation (LFIRE). A major difficulty in all these methods is choosing summary statistics that reduce the dimensionality of the data to facilitate inference. While several methods for choosing summary statistics have been proposed for ABC, the literature for synthetic likelihood and LFIRE is very thin to date. We here address this gap in the literature, focusing on the important special case of time-series models. We show that convolutional neural networks trained to predict the input parameters from the data provide suitable summary statistics for LFIRE. On a wide range of time-series models, a single neural network architecture produced equally or more accurate posteriors than alternative methods.

arXiv commentFor a demo, see https://traiko.com/pages/research/lfire/

References in corpus (4)

• Approximate Bayesian computation scheme for parameter inference and model selection in dynamical systems
• A Comparative Review of Dimension Reduction Methods in Approximate Bayesian Computation
• Approximating Likelihood Ratios with Calibrated Discriminative Classifiers
• Approximate Bayesian Computation for Astronomical Model Analysis: A Case Study in Galaxy Demographics and Morphological Transformation at High Redshift

Cited by in corpus (10)

• The frontier of simulation-based inference
• Mining gold from implicit models to improve likelihood-free inference
• Mining for Dark Matter Substructure: Inferring subhalo population properties from strong lenses with machine learning
• Likelihood-free MCMC with Amortized Approximate Ratio Estimators
• Benchmarking Simulation-Based Inference
• Neural Approximate Sufficient Statistics for Implicit Models
• Marginally-calibrated deep distributional regression
• Telescoping Density-Ratio Estimation
• Confidence Sets and Hypothesis Testing in a Likelihood-Free Inference Setting
• Detecting conflicting summary statistics in likelihood-free inference