Summary-powered prediction under distribution shift
Abstract
Prediction models can perform poorly when the deployment population differs from the training population. Data from the target population would help, but individual-level target data may be inaccessible because of access restrictions or reporting conventions. We consider a multi-resolution setting in which individual-level data are available from a source population, while the target population is observed only through subgroup summaries. We propose SAGE, a one-step estimator that updates a source-trained predictor using a gradient estimated from these summaries. Motivated by diagnostics consistent with the random distribution shift model, we choose SAGE's step-size to account for both sampling and distributional uncertainty. Under this model, SAGE reduces mean asymptotic target excess risk relative to the source-trained predictor. We also show that, under the model, entropy-balancing weighted empirical risk minimization (EB), which reweights source observations to match the target summaries, is asymptotically equivalent to a full-step SAGE update. SAGE with the optimal step-size has asymptotic mean squared error no larger than that of EB. Across real-world datasets, SAGE generally improves on the source-trained predictor and one-step updates that ignore distribution shift, including when the random shift model only partially captures the observed shifts. Compared to EB, SAGE improves prediction more consistently across the sample size and shift settings studied.
Related
- Beyond Point Predictions: Distribution-Informed Prediction-Powered Inference
- Missingness-Aware Conformal Prediction Under Cross-Hospital Distribution Shift
- PPI++: Efficient prediction-powered inference
- Covariance-Calibrated Prediction-Powered Inference
- Which Metrics Save the Most Human Annotation? Prediction-Powered Evaluation and Meta-Evaluation
BibTeX
@misc{zhang2026summarypowered,
title = {{Summary-powered prediction under distribution shift}},
author = {Ivy Zhang and Dominik Rothenhaeusler},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.30908},
eprint = {2609.30908},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.30908},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Summary-powered prediction under distribution shift
AU - Ivy Zhang
AU - Dominik Rothenhaeusler
PY - 2026
DA - 2026/09/25
PB - arXiv
DO - 10.48550/arxiv.2609.30908
UR - https://arxiv.org/abs/2609.30908
AB - Prediction models can perform poorly when the deployment population differs from the training population. Data from the target population would help, but individual-level target data may be inaccessible because of access restrictions or reporting conventions. We consider a multi-resolution setting in which individual-level data are available from a source population, while the target population is observed only through subgroup summaries. We propose SAGE, a one-step estimator that updates a source-trained predictor using a gradient estimated from these summaries. Motivated by diagnostics consistent with the random distribution shift model, we choose SAGE's step-size to account for both sampling and distributional uncertainty. Under this model, SAGE reduces mean asymptotic target excess risk relative to the source-trained predictor. We also show that, under the model, entropy-balancing weighted empirical risk minimization (EB), which reweights source observations to match the target summaries, is asymptotically equivalent to a full-step SAGE update. SAGE with the optimal step-size has asymptotic mean squared error no larger than that of EB. Across real-world datasets, SAGE generally improves on the source-trained predictor and one-step updates that ignore distribution shift, including when the random shift model only partially captures the observed shifts. Compared to EB, SAGE improves prediction more consistently across the sample size and shift settings studied.
ER -