Publications

Citation

BibTex format

@article{Alvey:2026:2632-2153/ae3103,
author = {Alvey, J and Contaldi, CR and Pieroni, M},
doi = {2632-2153/ae3103},
journal = {Machine Learning Science and Technology},
title = {Simulation-based inference with deep ensembles: evaluating calibration uncertainty and detecting model misspecification},
url = {http://dx.doi.org/10.1088/2632-2153/ae3103},
volume = {7},
year = {2026}
}
Download

RIS format (EndNote, RefMan)

TY  - JOUR
AB  - Simulation-based inference (SBI) offers a principled and flexible framework for conducting Bayesian inference in any situation where forward simulations are feasible. However, validating the accuracy and reliability of the inferred posteriors remains a persistent challenge. In this work, we point out a simple diagnostic approach rooted in ensemble learning methods to assess the internal consistency of SBI outputs that does not require access to the true posterior. By training multiple neural estimators under identical conditions and evaluating their pairwise Kullback–Leibler (KL) divergences, we define a consistency criterion that quantifies agreement across the ensemble. We highlight two core use cases for this framework: (a) for generating a robust estimate of the systematic uncertainty in parameter reconstruction associated with the training procedure, and (b) for detecting possible model misspecification when using trained estimators on real data. We also demonstrate the relationship between significant KL divergences and issues such as insufficient convergence due to, e.g. too low a simulation budget, or intrinsic variance in the training process. Overall, this ensemble-based diagnostic framework provides a lightweight, scalable, and model-agnostic tool for enhancing the trustworthiness of SBI in scientific applications.
AU  - Alvey,J
AU  - Contaldi,CR
AU  - Pieroni,M
DO  - 2632-2153/ae3103
PY  - 2026///
TI  - Simulation-based inference with deep ensembles: evaluating calibration uncertainty and detecting model misspecification
T2  - Machine Learning Science and Technology
UR  - http://dx.doi.org/10.1088/2632-2153/ae3103
VL  - 7
ER  -
Download

Note to staff:  Adding new publications to a research group

  1. Log in to Symplectic.
  2. Click on Menu > Create Links
  3. Choose what you want to create links between – in this case ‘Publications’ and ‘Organisational structures’.
  4. Choose the organisational structure (research group) into which you want to link the publications and check the box next to it.
  5. Now check the box of any publication you want to add to that group. You can use the filters to find what you want and select multiple publications if necessary. 
  6. Scroll to the bottom and click the blue ‘Create new link’ button to link them.
  7. The publications will be added to the group, and will be displayed on the group publications feed within 24 hours (it is not immediate).

Any problems, talk to Tim Evans or the Faculty Web Team.