Representative, Informative, and De-Amplifying: Requirements for Robust Bayesian Active Learning under Model Misspecification
Not on the map. Why?
Abstract (arXiv)
In many science and industry settings, a central challenge is designing experiments under time and budget constraints. Bayesian Optimal Experimental Design (BOED) is a paradigm to pick maximally informative designs that has been widely applied to such problems. During training, BOED selects inputs according to a pre-determined acquisition criterion to target informativeness. During testing, the model learned during training encounters a naturally occurring distribution of test samples. This leads to an instance of covariate shift, where the train and test samples are drawn from different distributions (the training samples are not representative of the test distribution). Prior work has shown that in the presence of model misspecification, covariate shift amplifies generalization error. Our first contribution is to provide a mathematical analysis of generalization error in the presence of model misspecification, revealing that, beyond covariate shift, generalization error is also driven by a previously unidentified phenomenon we term error (de-)amplification. We then develop a new acquisition function that mitigates the effects of model misspecification by including terms for representativeness, informativeness, and de-amplification (R-IDeA). Our experimental results demonstrate that the proposed method performs better than methods that target only informativeness, only representativeness, or both.
Key facts. Cited by 2 papers in the map (data through 14 Oct 2026). Builds on 28 of its 79 references that are in the map. First posted on arXiv 9 Jun 2025.
Builds on
28 of 79 references.
References · 28
- 2019PyTorch: An Imperative Style, High-Performance Deep Learning Librarycited by 20,305
- 2021Convex optimizationcited by 2,515
- 2018Pyro: Deep Universal Probabilistic Programmingcited by 380
- 2023Modern Bayesian Experimental Designcited by 125
- 2019Variational Bayesian Optimal Experimental Designcited by 116
- 2021Deep Adaptive Design: Amortizing Sequential Bayesian Experimental Designcited by 106
- 2017Learning Active Learning from Datacited by 101
- 2014Inconsistency of Bayesian Inference for Misspecified Linear Models, and a Proposal for Repairing Itcited by 95
- 2021On Statistical Bias In Active Learning: How and When To Fix Itcited by 75
- 2021Detecting Model Misspecification in Amortized Bayesian Inference with Neural Networkscited by 49
- 2023Learning Robust Statistics for Simulation-based Inference under Model Misspecificationcited by 48
- 2023Credal Bayesian Deep Learningcited by 45
16 more
- 2024Optimal experimental design: Formulations and computationscited by 45
- 2023Maximum Likelihood Estimation is All You Need for Well-Specified Covariate Shiftcited by 15
- 2023Optimally-Weighted Estimators of the Maximum Mean Discrepancy for Likelihood-Free Inferencecited by 13
- 2024Variational Bayesian Optimal Experimental Design with Normalizing Flowscited by 13
- 2022Bayesian optimization with known experimental and design constraints for chemistry applicationscited by 11
- 2023Bayesian inference for misspecified generative modelscited by 11
- 2025Simulation-based Bayesian inference under model misspecificationcited by 9
- 2019Bayesian decision-theoretic design of experiments under an alternative modelcited by 8
- 2022Characterizing the robustness of Bayesian adaptive experimental designs to active learning biascited by 8
- 2023Metrics for Bayesian Optimal Experiment Design under Model Misspecificationcited by 7
- 2023Bayesian Active Learning in the Presence of Nuisance Parameterscited by 6
- 2023Reliable Bayesian Inference in Misspecified Modelscited by 4
- 2024Mitigating Covariate Shift in Misspecified Regression with Applications to Reinforcement Learningcited by 4
- 2018Minimizing Sensitivity to Model Misspecificationcited by 3
- 2021Partial Wasserstein and Maximum Mean Discrepancy distances for bridging the gap between outlier detection and drift detectioncited by 3
- 2023Gibbs optimal design of experimentscited by 2
Built on by
Citing papers per year: 2026: 2.
2 papers.