Diffusion Models in Simulation-Based Inference: A Tutorial Review

Jonas Arruda, Niels Bracher, Ullrich Köthe, Jan Hasenauer, Stefan T. Radev · 2025

Read Diffusion Models in Simulation-Based Inference: A Tutorial Review by Jonas Arruda, Niels Bracher, Ullrich Köthe, Jan Hasenauer, Stefan T. Radev (2025) 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-2512-20685.txt.

Introduction

Jonas Arruda and co-authors review, in tutorial form, how diffusion models can serve simulation-based inference. In many fields, a model can be run to produce data for any chosen settings, yet no formula gives the chance of real data. Inference based on simulation turns the model runs into odds over the unknown settings. Diffusion models have lately become strong engines for this job, since they can learn rich shapes over settings and data alike.

The review sorts out the choices in training, use, and testing such models. It covers guidance, score mixing, flow matching, consistency models, and joint models, and shows how noise schedules and samplers trade speed against accuracy. Case studies range over problem sizes, budgets for model runs, and model types. Open questions end the tour, marking where this young field still needs work.

Catalog record

CREATORJonas Arruda, Niels Bracher, Ullrich Köthe, Jan Hasenauer, Stefan T. Radev
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE193 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2512-20685

Browse the full library · About the Tower of Babel