Fairness under Covariate Shift: Improving Fairness-Accuracy Tradeoff with Few Unlabeled Test Samples
Shreyas Havaldar, Jatin Chauhan, Karthikeyan Shanmugam, Jay Nandy, Aravindan Raghuveer
Abstract
Covariate shift in the test data is a common practical phenomena that can significantly downgrade both the accuracy and the fairness performance of the model. Ensuring fairness across different sensitive groups under covariate shift is of paramount importance due to societal implications like criminal justice. We operate in the unsupervised regime where only a small set of unlabeled test samples along with a labeled training set is available. Towards improving fairness under this highly challenging yet realistic scenario, we make three contributions. First is a novel composite weighted entropy based objective for prediction accuracy which is optimized along with a representation matching loss for fairness. We experimentally verify that optimizing with our loss formulation outperforms a number of state-of-the-art baselines in the pareto sense with respect to the fairness-accuracy tradeoff on several standard datasets. Our second contribution is a new setting we term Asymmetric Covariate Shift that, to the best of our knowledge, has not been studied before. Asymmetric covariate shift occurs when distribution of covariates of one group shifts significantly compared to the other groups and this happens when a dominant group is over-represented. While this setting is extremely challenging for current baselines, We show that our proposed method significantly outperforms them. Our third contribution is theoretical, where we show that our weighted entropy term along with prediction loss on the training set approximates test loss under covariate shift. Empirically and through formal sample complexity bounds, we show that this approximation to the unseen test loss does not depend on importance sampling variance which affects many other baselines.
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
Lineage
- Tent: Fully Test-Time Adaptation by Entropy Minimization cites Improving robustness against common corruptions by covariate shift adaptation
- A Fine-Grained Analysis on Distribution Shift cites Improving robustness against common corruptions by covariate shift adaptation
Builds on9
- Tent: Fully Test-Time Adaptation by Entropy MinimizationDequan Wang, Evan Shelhamer, Shaoteng Liu, Bruno A. Olshausen et al.ICLR 2021 · 1,731 citations
- Test-Time Training with Self-Supervision for Generalization under Distribution ShiftsYu Sun, Xiaolong Wang, Zhuang Liu, John Miller et al.ICML 2020 · 1,220 citations
- Improving robustness against common corruptions by covariate shift adaptationSteffen Schneider, Evgenia Rusak, Luisa Eck, Oliver Bringmann et al.NeurIPS 2020 · 688 citations
- Gradient Matching for Domain GeneralizationYuge Shi, Jeffrey Seely, Philip H. S. Torr, Siddharth Narayanaswamy et al.ICLR 2022 · 358 citations
- A Fine-Grained Analysis on Distribution ShiftOlivia Wiles, Sven Gowal, Florian Stimberg, Sylvestre-Alvise Rebuffi et al.ICLR 2022 · 258 citations
Related papers
- Robust Fairness Under Covariate ShiftAshkan Rezaei, Anqi Liu, Omid Memarrast, Brian D. ZiebartAAAI 2021 · 94 citations
- Self-Supervised Fair Representation Learning without DemographicsJunyi Chai, Xiaoqian WangNeurIPS 2022 · 35 citations
- Positive-unlabeled AUC Maximization under Covariate ShiftAtsutoshi Kumagai, Tomoharu Iwata, Hiroshi Takahashi, Taishi Nishiyama et al.ICML 2025
- Fairness with Adaptive WeightsJunyi Chai, Xiaoqian WangICML 2022 · 47 citations
- Fairness Transferability Subject to Bounded Distribution ShiftYatong Chen, Reilly Raab, Jialu Wang, Yang LiuNeurIPS 2022 · 40 citations
