arXiv:2607. 08335v1 Announce Type: cross Abstract: Policy-based approaches to Bayesian experimental design (BED) allow the learning of deep policy networks that adaptively make intelligent design decisions based on previously collected data.
The paper introduces Generalised Bayesian Optimal Experimental Design (GBOED), extending generalised Bayesian inference to experimental design. It replaces the likelihood in Bayesian updates with a loss function, deriving a new acquisition function called Gibbs expected information gain. Empirical results show that GBOED improves robustness against outliers and mis-specified outcome noise distributions.
arXiv:2502. 08004v2 Announce Type: replace-cross Abstract: Simulation-based inference (SBI) is a method to perform inference on a variety of complex scientific models with challenging inference (inverse) problems.
arXiv:2512. 22999v2 Announce Type: replace-cross Abstract: We consider problems of parameter estimation where design variables can be actively optimized to maximize information gain.
arXiv:2607. 16927v1 Announce Type: cross Abstract: We introduce Deep Adaptive Bayesian Screening (DABS), a method for performing adaptive factorial screening in high-dimensional discrete design spaces.
The paper introduces a learning-based surrogate approach for stochastic optimization problems where uncertainty depends on the decision, modeled via a nonparametric regression. It constructs a surrogate that embeds iteratively updated Jacobian estimates, using an adaptive random design that focuses sampling near the current iterate to achieve dimension‑independent convergence of the Jacobian estimates. The resulting learning‑based stochastic prox‑linear (L‑SPL) algorithm demonstrates nonasymptotic convergence rates and outperforms existing methods in sample efficiency and objective value in numerical experiments.
The paper proposes a method to embed sequential Bayesian experimental design (BED) into large language models (LLMs) by training the models to reason about action selection using information-gain rewards. By specifying the experimental generative model separately from the LLM’s beliefs, the authors optimize expected information gain under a chosen model. Experiments on the Twenty Questions task show that the resulting policies outperform both inference- and training-time BED baselines while incurring lower computational costs, and the approach generalizes to unseen categories and transfers positively to the Medi-Q clinical dialogue benchmark.
The paper explores how data from fixed A/B tests can guide the deployment of adaptive experiments using contextual bandits. By combining off‑policy evaluation with a controlled warm‑start simulation, the authors rank pre‑specified adaptive and non‑adaptive policies using doubly robust estimators. Experiments on synthetic trials and real benchmarks show that adaptive, context‑aware policies outperform fixed allocations when heterogeneity exists, but offer little advantage otherwise.
The paper presents a method for approximating differential entropy in Bayesian optimal experimental design, focusing on cases where the likelihood entropy is design‑independent or explicitly computable, such as additive noise models. By reducing the expected information gain to a single integral, the authors propose using Monte Carlo or quasi‑Monte Carlo surrogates for the evidence density while evaluating differential entropy with standard techniques, thereby avoiding extra likelihood evaluations. Theoretical analysis shows convergence rates comparable to or better than existing methods, and numerical experiments support these findings.
arXiv:2406.04261v2 Announce Type: replace Abstract: Simulators are vital in science and engineering, as they faithfully model the influence of design parameters on real-world observations. Optimizing...
arXiv:2606. 25197v1 Announce Type: new Abstract: Learning effective policies for adaptive data acquisition remains challenging: posterior-based methods rely on surrogate models and posterior approximations that can be misspecified or biased, while direct policy-learning methods map from historical observations and fail to exploit available model representations, making learning harder.
The paper introduces usage‑aware experimental design, a framework that extends standard Bayesian experimental design to settings where external decision makers act on collected data. It trains adaptive design policies to minimise the experimenter’s expected loss under external rules, which may be non‑Bayesian or black‑box. The authors show that more informative data can be harmful under such rules, so usage risk cannot be expressed as expected uncertainty reduction, thereby generalising standard BED.