Model Informed Flows for Bayesian Inference of Probabilistic Programs

Joohwan Ko, Justin Domke · 2025

Read Model Informed Flows for Bayesian Inference of Probabilistic Programs by Joohwan Ko, Justin Domke (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-2505-24243.txt.

Description

Joohwan Ko and Justin Domke present Model Informed Flows for Bayesian inference in this 2025 paper. Variational inference approximates a posterior distribution with a simpler family. Complex layered models create posteriors with awkward shapes that simple families fit poorly. Flow based families and a reparametrization trick each help in part.

The authors prove the two tricks together equal a certain autoregressive flow with an added shift. That insight guides a new architecture. It adds the shift mechanism, prior information, and the model layer order. Tests on layered and flat benchmark models gave tighter posterior fits. Performance matched or beat the best prior methods. The work turns a theoretical link into a practical inference tool.

Catalog record

CREATORJoohwan Ko, Justin Domke
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE54 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2505-24243

Browse the full library · About the Tower of Babel