Amortized Simulation-Based Inference in Generalized Bayes via Neural Posterior Estimation

Shiyi Sun, Geoff K. Nicholls, Jeong Eun Lee · 2026

Read Amortized Simulation-Based Inference in Generalized Bayes via Neural Posterior Estimation by Shiyi Sun, Geoff K. Nicholls, Jeong Eun Lee (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2601-22367.txt.

Description

Shiyi Sun, Geoff Nicholls, and Jeong Eun Lee speed up a safe form of Bayesian inference. The generalized form softens the loss with a heat-like dial, which guards against a wrong model. Their 2026 preprint notes that old samplers for it are slow and must restart for each new data set and dial value.

Their method trains a neural net to give the answer at once, for any dial value. Fast, safe inference helps when models are known to be rough. It fits fields from climate to health. The work makes a careful method cheap enough for daily use.

Catalog record

CREATORShiyi Sun, Geoff K. Nicholls, Jeong Eun Lee
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE44 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2601-22367

Browse the full library · About the Tower of Babel