Step-DAD: Semi-Amortized Policy-Based Bayesian Experimental Design

Abstract

We develop a semi-amortized, policy-based, approach to Bayesian experimental design (BED) called Stepwise Deep Adaptive Design (Step-DAD). Like existing, fully amortized, policy-based BED approaches, Step-DAD trains a design policy upfront before the experiment. However, rather than keeping this policy fixed, Step-DAD periodically updates it as data is gathered, refining it to the particular experimental instance. This test-time adaptation improves both the flexibility and the robustness of the design strategy compared with existing approaches. Empirically, Step-DAD consistently demonstrates superior decision-making and robustness compared with current state-of-the-art BED methods.

Cite

Text

Hedman et al. "Step-DAD: Semi-Amortized Policy-Based Bayesian Experimental Design." Proceedings of the 42nd International Conference on Machine Learning, 2025.

Markdown

[Hedman et al. "Step-DAD: Semi-Amortized Policy-Based Bayesian Experimental Design." Proceedings of the 42nd International Conference on Machine Learning, 2025.](https://mlanthology.org/icml/2025/hedman2025icml-stepdad/)

BibTeX

@inproceedings{hedman2025icml-stepdad,
  title     = {{Step-DAD: Semi-Amortized Policy-Based Bayesian Experimental Design}},
  author    = {Hedman, Marcel and Ivanova, Desi R. and Guan, Cong and Rainforth, Tom},
  booktitle = {Proceedings of the 42nd International Conference on Machine Learning},
  year      = {2025},
  pages     = {22904-22923},
  volume    = {267},
  url       = {https://mlanthology.org/icml/2025/hedman2025icml-stepdad/}
}