JADAI: Jointly Amortizing Adaptive Design and Bayesian Inference

Niels Leif Bracher, Lars Kühmichel, Desi R. Ivanova, Xavier Intes, Paul-Christian Bürkner, Stefan T. Radev
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:9467-9493, 2026.

Abstract

We consider problems of parameter estimation where design variables can be actively optimized to maximize information gain. To this end, we introduce JADAI, a framework that jointly amortizes Bayesian adaptive design and inference by training a policy, a history network, and an inference network end-to-end. The networks minimize a generic loss that aggregates incremental reductions in posterior error along experimental sequences without density evaluations. Inference networks are instantiated with diffusion models that can approximate high-dimensional and multimodal posteriors at every experimental step. JADAI achieves superior or competitive performance across adaptive design benchmarks.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-bracher26a, title = {{JADAI}: Jointly Amortizing Adaptive Design and {B}ayesian Inference}, author = {Bracher, Niels Leif and K\"{u}hmichel, Lars and Ivanova, Desi R. and Intes, Xavier and B\"{u}rkner, Paul-Christian and Radev, Stefan T.}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {9467--9493}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/bracher26a/bracher26a.pdf}, url = {https://proceedings.mlr.press/v306/bracher26a.html}, abstract = {We consider problems of parameter estimation where design variables can be actively optimized to maximize information gain. To this end, we introduce JADAI, a framework that jointly amortizes Bayesian adaptive design and inference by training a policy, a history network, and an inference network end-to-end. The networks minimize a generic loss that aggregates incremental reductions in posterior error along experimental sequences without density evaluations. Inference networks are instantiated with diffusion models that can approximate high-dimensional and multimodal posteriors at every experimental step. JADAI achieves superior or competitive performance across adaptive design benchmarks.} }
Endnote
%0 Conference Paper %T JADAI: Jointly Amortizing Adaptive Design and Bayesian Inference %A Niels Leif Bracher %A Lars Kühmichel %A Desi R. Ivanova %A Xavier Intes %A Paul-Christian Bürkner %A Stefan T. Radev %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-bracher26a %I PMLR %P 9467--9493 %U https://proceedings.mlr.press/v306/bracher26a.html %V 306 %X We consider problems of parameter estimation where design variables can be actively optimized to maximize information gain. To this end, we introduce JADAI, a framework that jointly amortizes Bayesian adaptive design and inference by training a policy, a history network, and an inference network end-to-end. The networks minimize a generic loss that aggregates incremental reductions in posterior error along experimental sequences without density evaluations. Inference networks are instantiated with diffusion models that can approximate high-dimensional and multimodal posteriors at every experimental step. JADAI achieves superior or competitive performance across adaptive design benchmarks.
APA
Bracher, N.L., Kühmichel, L., Ivanova, D.R., Intes, X., Bürkner, P. & Radev, S.T.. (2026). JADAI: Jointly Amortizing Adaptive Design and Bayesian Inference. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:9467-9493 Available from https://proceedings.mlr.press/v306/bracher26a.html.

Related Material