Rural Readiness: Domain Generalization with Adversarially Chosen Domains
Abstract
Predictive models trained at well-resourced data-rich institutions often fail when deployed in new populations. Motivated by the goal of building clinical diagnosis and decision-making algorithms for data-scarce rural populations using the general knowledge learned from data-rich hospitals, we formulate the fundamental question of \emph{domain generalization with adversarially chosen domains}, which asks a learner to perform on any unseen target domain nearly as well as the worst hypothesis consistent with the evidence from the observed source domains. Assuming only that the hypothesis class contains a "universal" predictor with small error on all k source domains, we characterize the sample complexity of this objective up to logarithmic factors. Letting d denote the VC dimension of the hypothesis class, we find, somewhat surprisingly, that the sample complexity reduces significantly from Ω(dk) to O(d) under a small relaxation of the learning goal in which, roughly speaking, we allow the learned model to compete with hypotheses that perform well on 99% instead of 100% of the source domains. By applying this insight to the meta-distributional setting, in which source domains are selected according to a distribution on distributions, we obtain a corresponding improvement over known sample complexity upper bounds. We also study the effect of data access to the target domain and show that a small labeled sample from the target domain does not improve the worst-case sample complexity on the source domains.
Related
- On the generalization of Christol's theorem to domains of finite type over finite fields
- ACOUSTIC DOMAIN SHIFT IN SPOKEN LANGUAGE IDENTIFICATION FROM SYSTEMATIC DOMAIN GENERALIZATION EVALUATION TO REAL-WORLD APPLICATION
- Chaos Is a LADDER: Domain Generalization Beyond Invariance via Reweighting
- Eigenfunction asymptotics and nodal domain estimates for dumbbell domains
- Persistent mixed worldviews in the theory of chosen preferences
BibTeX
@misc{dwork2026rural,
title = {{Rural Readiness: Domain Generalization with Adversarially Chosen Domains}},
author = {Cynthia Dwork and Andisheh Ghasemi and Lunjia Hu},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.40468},
eprint = {2610.40468},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.40468},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Rural Readiness: Domain Generalization with Adversarially Chosen Domains
AU - Cynthia Dwork
AU - Andisheh Ghasemi
AU - Lunjia Hu
PY - 2026
DA - 2026/10/29
PB - arXiv
DO - 10.48550/arxiv.2610.40468
UR - https://arxiv.org/abs/2610.40468
AB - Predictive models trained at well-resourced data-rich institutions often fail when deployed in new populations. Motivated by the goal of building clinical diagnosis and decision-making algorithms for data-scarce rural populations using the general knowledge learned from data-rich hospitals, we formulate the fundamental question of \emph{domain generalization with adversarially chosen domains}, which asks a learner to perform on any unseen target domain nearly as well as the worst hypothesis consistent with the evidence from the observed source domains. Assuming only that the hypothesis class contains a "universal" predictor with small error on all k source domains, we characterize the sample complexity of this objective up to logarithmic factors. Letting d denote the VC dimension of the hypothesis class, we find, somewhat surprisingly, that the sample complexity reduces significantly from Ω(dk) to O(d) under a small relaxation of the learning goal in which, roughly speaking, we allow the learned model to compete with hypotheses that perform well on 99% instead of 100% of the source domains. By applying this insight to the meta-distributional setting, in which source domains are selected according to a distribution on distributions, we obtain a corresponding improvement over known sample complexity upper bounds. We also study the effect of data access to the target domain and show that a small labeled sample from the target domain does not improve the worst-case sample complexity on the source domains.
ER -