Modular Bayesian Inference for Partially Misspecified Mixtures with Uncertain Class Labels
Abstract
Scientific inference often involves observations from multiple latent populations whose class memberships are uncertain and must be inferred from available data. While full Bayesian mixture modelling provides a principled approach to joint uncertainty quantification in this setting, often the model is misspecified or less trustworthy for populations that are not the target of the inference, which can distort posterior inference for the parameters of interest. We propose a modular Bayesian framework that safeguards against such biases. We decouple classification from downstream inference using fixed class probabilities, for example from a classifier trained on a subset of labelled data. We then protect inference on the parameters of interest by preventing the likelihood contributions of less trustworthy populations from influencing them. A key benefit of this principled modularization is that it greatly simplifies inference, reducing a fully joint mixture analysis to sampling label allocations, fitting the protected downstream model for each sample, and averaging the resulting posteriors. We establish calibration conditions on these class probabilities under which the proposed procedure preserves the inferential target of the analysis with known labels, and demonstrate the method in synthetic experiments and in cosmological analysis of photometric supernova samples.
Related
- Solar spicule population properties inferred from IRIS Mg II k off-limb spectra using likelihood-free Bayesian inference
- Fast Bayesian Inference for Long-Duration Gravitational-Wave Signals in 3G detectors
- A space of inference spaces in the space sciences - parametric Bayesian inference in astronomy, cosmology and particle physics
- VROOM-SBI: A Fast Simulation-Based Bayesian Inference Methodology for Stokes QU-Fitting in Radio Interferometry
- MUSHY: Multimodal Flow-Based Amortized Bayesian Inference for Spectroscopic Data Fusion
BibTeX
@misc{autenrieth2026modular,
title = {{Modular Bayesian Inference for Partially Misspecified Mixtures with Uncertain Class Labels}},
author = {Maximilian Autenrieth and Konstantin Karchev and David A. van Dyk and Kaisey Mandel and Roberto Trotta},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.27085},
eprint = {2610.27085},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.27085},
}
RIS
TY - UNPB
TI - Modular Bayesian Inference for Partially Misspecified Mixtures with Uncertain Class Labels
AU - Maximilian Autenrieth
AU - Konstantin Karchev
AU - David A. van Dyk
AU - Kaisey Mandel
AU - Roberto Trotta
PY - 2026
DA - 2026/10/19
PB - arXiv
DO - 10.48550/arxiv.2610.27085
UR - https://arxiv.org/abs/2610.27085
AB - Scientific inference often involves observations from multiple latent populations whose class memberships are uncertain and must be inferred from available data. While full Bayesian mixture modelling provides a principled approach to joint uncertainty quantification in this setting, often the model is misspecified or less trustworthy for populations that are not the target of the inference, which can distort posterior inference for the parameters of interest. We propose a modular Bayesian framework that safeguards against such biases. We decouple classification from downstream inference using fixed class probabilities, for example from a classifier trained on a subset of labelled data. We then protect inference on the parameters of interest by preventing the likelihood contributions of less trustworthy populations from influencing them. A key benefit of this principled modularization is that it greatly simplifies inference, reducing a fully joint mixture analysis to sampling label allocations, fitting the protected downstream model for each sample, and averaging the resulting posteriors. We establish calibration conditions on these class probabilities under which the proposed procedure preserves the inferential target of the analysis with known labels, and demonstrate the method in synthetic experiments and in cosmological analysis of photometric supernova samples.
ER -