2512.22999

Total: 1

#1 JADAI: Jointly Amortizing Adaptive Design and Bayesian Inference [PDF] [Copy] [Kimi] [REL]

Authors: Niels Bracher, Lars Kühmichel, Desi R. Ivanova, Xavier Intes, Paul-Christian Bürkner, Stefan T. Radev

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. Inference networks are instantiated with diffusion-based posterior estimators that can approximate high-dimensional and multimodal posteriors at every experimental step. Across standard adaptive design benchmarks, JADAI achieves superior or competitive performance.

Subjects: Machine Learning , Artificial Intelligence , Machine Learning

Publish: 2025-12-28 16:54:43 UTC