Latent-Space Importance Weighting under Covariate Shift
Abstract
In this work, we propose a transfer learning algorithm for covariate shift, integrating sample reweighting with representation learning. A key feature of our approach is the use of a lower-dimensional latent space for modeling, in contrast to conventional methods that operate directly in the high-dimensional ambient space. Our method estimates the density ratio of learned latent features for downstream weighted prediction. We establish explicit target-domain excess-risk rates with independent data folds for strictly sufficient learned representations with regular learned-space prediction scores and density ratios. We give dimension comparisons under stated smoothness conditions and a variance comparison for true weights and a fixed predictor. General finite-sample inequalities retain representation-quality errors. Experiments on simulated and real-world datasets demonstrate gains from combining learned representations and latent weighting, and reveal sensitivity to initialization and domain choice.
Related
- Importance Weighting for Unlabeled-unlabeled Learning under Distribution Shift
- LUCID: Local Understanding of Covariate Importance with Conditional Density Estimation
- Ensembles of Random Partition-Based Models under Covariate Shift
- Conformalized Quantile Regression and Minimax Limits of Fixed-Score Calibration under Known Covariate Shift
- Backward Conformal Prediction under Covariate Shift
BibTeX
@misc{jiao2026latentspace,
title = {{Latent-Space Importance Weighting under Covariate Shift}},
author = {Yuling Jiao and Lican Kang and Yang Xiang and Jerry Zhijian Yang and Fusheng Zhou},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.15571},
eprint = {2610.15571},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.15571},
}
RIS
TY - UNPB
TI - Latent-Space Importance Weighting under Covariate Shift
AU - Yuling Jiao
AU - Lican Kang
AU - Yang Xiang
AU - Jerry Zhijian Yang
AU - Fusheng Zhou
PY - 2026
DA - 2026/10/23
PB - arXiv
DO - 10.48550/arxiv.2610.15571
UR - https://arxiv.org/abs/2610.15571
AB - In this work, we propose a transfer learning algorithm for covariate shift, integrating sample reweighting with representation learning. A key feature of our approach is the use of a lower-dimensional latent space for modeling, in contrast to conventional methods that operate directly in the high-dimensional ambient space. Our method estimates the density ratio of learned latent features for downstream weighted prediction. We establish explicit target-domain excess-risk rates with independent data folds for strictly sufficient learned representations with regular learned-space prediction scores and density ratios. We give dimension comparisons under stated smoothness conditions and a variance comparison for true weights and a fixed predictor. General finite-sample inequalities retain representation-quality errors. Experiments on simulated and real-world datasets demonstrate gains from combining learned representations and latent weighting, and reveal sensitivity to initialization and domain choice.
ER -