The frontier of simulation-based inference

By Kyle Cranmer, J. Brehmer, Gilles Louppe. 2019

1,284 citations recorded by the source corpus.

Many domains of science have developed complex simulations to describe phenomena of interest. While these simulations provide high-fidelity models, they are poorly suited for inference and lead to challenging inverse problems. We review the rapidly developing field of simulation-based inference and identify the forces giving additional momentum to the field. Finally, we describe how the frontier is expanding so that a broad audience can appreciate the profound influence these developments may have on science.

Subfields: Artificial Intelligence, Statistical and Nonlinear Physics, Probabilistic & Bayesian Methods

How researchers see this paper

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

In Artificial Intelligence

In Statistical and Nonlinear Physics

In Probabilistic & Bayesian Methods

Citations in the corpus

Language Model Cascades (2022, methodology): “Then techniques from simulation based inference, for example, can be applied to do inference in such situations (Cranmer et al., 2020).”
Normalizing Flows for Probabilistic Modeling and Inference (2019, methodology): “Inferring the parameters η of a simulator-based model given observed data x is often referred to as likelihood-free inference (Papamakarios, 2019; Cranmer et al., 2019).”
On scientific understanding with artificial intelligence (2022, background): “, combined with methods such as causal inference [98], can be improved to propose reasonable physical models of unknown systems that advance scientific understanding.”
sbi: A toolkit for simulation-based inference (2020, methodology): “Traditional algorithms for this ‘likelihood-free’ setting (Cranmer, Brehmer, & Louppe, 2020) are based on Monte-Carlo rejection (Pritchard, Seielstad, Perez-Lezaun, & Feldman, 1999; Sisson, Fan, & Tanaka, 2007), an approach known as Approximate Bayesian Computation (ABC).”
BayesFlow: Learning Complex Stochastic Models With Invertible Neural Networks (2020, background): “More details can be found in the section Related Work and in the excellent review by [9].”
Meta-Sim2: Unsupervised Learning of Scene Structure for Synthetic Data Generation (2020, background): “[11] provide a great overview of advances in simulator based inference.”
Transformers Can Do Bayesian Inference (2021, background): “One more line of research to be mentioned is amortized simulation-based inference (Cranmer et al., 2020; Chan et al., 2018), in particular neural posterior estimation (Lueckmann et al.”