Likelihood-free MCMC with Amortized Approximate Ratio Estimators

270 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

The frontier of simulation-based inference (2019, methodology): “In most cases, however, we recommend trying methods based on training a neural network surrogate for the likelihood (41, 69) or the likelihood ratio (57, 73, 76).”
Towards constraining warm dark matter with stellar streams through neural simulation-based inference (2020, methodology): “(3) Hermans et al. (2019) show that an amortised estimator 𝑟 (𝑥 |𝜗) of the intractable likelihood-to-evidence ratio can be obtained by training a discriminator 𝑑 (𝜗, 𝑥) with inputs 𝜗 and 𝑥, to distinguish between samples from the joint 𝑝(𝜗, 𝑥) with class label 1 and samples from the…”
Likelihood-Free Inference by Ratio Estimation (2016, methodology): “For instance, Dinev and Gutmann (2018) tailored the approach to the special case of time-series models, Rogers-Smith et al. (2018) adapted the framework to a sequential population Monte Carlo scheme with adaptive proposals, and Hermans et al. (2020) perform (sequential) LFIRE for amortised likelihoo”
Mining for Dark Matter Substructure: Inferring Subhalo Population Properties from Strong Lenses with Machine Learning (2019, background): “…et al. 2017; Louppe & Cranmer 2017; Lueckmann et al. 2017; Papamakarios et al. 2017; Tran et al. 2017; Chen et al. 2018; Dinev & Gutmann 2018; Grathwohl et al. 2018; Huang et al. 2018; Kingma & Dhariwal 2018; Lueckmann et al. 2018; Papamakarios et al. 2018; Alsing et al. 2019; Hermans et al. 2019).”