Sequential Neural

Sequential Neural Posterior Estimation (SNPE) is a machine learning approach to Bayesian inference for models with intractable likelihood functions, aiming to efficiently estimate posterior distributions of model parameters using only simulated data. Current research focuses on improving SNPE's efficiency and robustness through techniques like adaptive calibration kernels, multi-level Monte Carlo methods, and the use of normalizing flows to handle high-dimensional data, often incorporating neural networks for density estimation. These advancements enable SNPE to tackle complex scientific problems, such as exoplanet atmospheric retrieval and Bayesian system identification, where traditional methods struggle due to computational cost or model complexity. The resulting speed and accuracy improvements are significantly impacting various fields by allowing for more sophisticated model usage and more reliable uncertainty quantification.

Papers

January 30, 2024

Leveraging Nested MLMC for Sequential Neural Posterior Estimation with Intractable Likelihoods
Xiliang Yang, Yifei Xiong, Zhijian He
Monte Carlo Conditional Density Intractable Likelihood Sequential Neural

January 8, 2024

FlopPITy: Enabling self-consistent exoplanet atmospheric retrievals with machine learning
Francisco Ardévol Martínez, Michiel Min, Daniela Huppenkothen, Inga Kamp, Paul I. Palmer
Machine Learning Exoplanet Candidate Atmospheric Retrieval Sequential Neural Brown Dwarf

November 28, 2023

Pseudo-Likelihood Inference
Theo Gruner, Boris Belousov, Fabio Muratore, Daniel Palenicek, Jan Peters
Simulation Based Inference Approximate Bayesian Computation Pseudo Likelihood Sequential Neural

November 21, 2023

An efficient likelihood-free Bayesian inference method based on sequential neural posterior estimation
Yifei Xiong, Xiliang Yang, Sanguo Zhang, Zhijian He
Loss Function Conditional Density Sequential Neural Likelihood Free

August 2, 2023

Simulation-based inference using surjective sequential neural likelihood estimation
Simon Dirmeier, Carlo Albert, Fernando Perez-Cruz
Simulation Based Inference Likelihood Based Surrogate Likelihood Sequential Neural

January 31, 2023

Misspecification-robust Sequential Neural Likelihood for Simulation-based Inference
Ryan P. Kelly, David J. Nott, David T. Frazier, David J. Warne, Chris Drovandi
Simulation Based Inference Model Misspecification Approximate Bayesian Computation Sequential Neural Synthetic Likelihood

October 10, 2022

Sequential Neural

Papers

Leveraging Nested MLMC for Sequential Neural Posterior Estimation with Intractable Likelihoods

FlopPITy: Enabling self-consistent exoplanet atmospheric retrievals with machine learning

Pseudo-Likelihood Inference

An efficient likelihood-free Bayesian inference method based on sequential neural posterior estimation

Simulation-based inference using surjective sequential neural likelihood estimation

Misspecification-robust Sequential Neural Likelihood for Simulation-based Inference

Sequential Neural Score Estimation: Likelihood-Free Inference with Conditional Score Based Diffusion Models

Truncated proposals for scalable and hassle-free simulation-based inference