Iterative Robust Satisficing: Minimizing Performance Degradation Under Distribution Shift

Enes Ağırman, Artun Saday, Cem Tekin
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:939-960, 2026.

Abstract

Modern neural networks often achieve high accuracy on their training distribution but degrade sharply under distribution shifts. We address this problem through Robust Satisficing (RS), an optimization objective that seeks parameters which attain a target level of in-distribution performance while minimizing fragility, defined as the rate at which performance deteriorates as the data distribution departs from training. We develop a gradient-based algorithm, Iterative Robust Satisficing (IRS), that directly optimizes this criterion. Across a range of synthetic and real-world distribution shifts, including long-tailed image classification, group shifts induced by spurious correlations, and natural shifts in tabular regression, IRS consistently improves performance on minority and worst-case groups without sacrificing overall accuracy. Notably, IRS achieves these robustness gains with a per-step computational cost similar to standard stochastic gradient descent and requires only a single forward and backward pass per update. Together, these results suggest that minimizing fragility provides a practical and effective alternative to existing robust training methods for learning models that remain reliable under distribution shift.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-agirman26a, title = {Iterative Robust Satisficing: Minimizing Performance Degradation Under Distribution Shift}, author = {A\u{g}{\i}rman, Enes and Saday, Artun and Tekin, Cem}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {939--960}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/agirman26a/agirman26a.pdf}, url = {https://proceedings.mlr.press/v306/agirman26a.html}, abstract = {Modern neural networks often achieve high accuracy on their training distribution but degrade sharply under distribution shifts. We address this problem through Robust Satisficing (RS), an optimization objective that seeks parameters which attain a target level of in-distribution performance while minimizing fragility, defined as the rate at which performance deteriorates as the data distribution departs from training. We develop a gradient-based algorithm, Iterative Robust Satisficing (IRS), that directly optimizes this criterion. Across a range of synthetic and real-world distribution shifts, including long-tailed image classification, group shifts induced by spurious correlations, and natural shifts in tabular regression, IRS consistently improves performance on minority and worst-case groups without sacrificing overall accuracy. Notably, IRS achieves these robustness gains with a per-step computational cost similar to standard stochastic gradient descent and requires only a single forward and backward pass per update. Together, these results suggest that minimizing fragility provides a practical and effective alternative to existing robust training methods for learning models that remain reliable under distribution shift.} }
Endnote
%0 Conference Paper %T Iterative Robust Satisficing: Minimizing Performance Degradation Under Distribution Shift %A Enes Ağırman %A Artun Saday %A Cem Tekin %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-agirman26a %I PMLR %P 939--960 %U https://proceedings.mlr.press/v306/agirman26a.html %V 306 %X Modern neural networks often achieve high accuracy on their training distribution but degrade sharply under distribution shifts. We address this problem through Robust Satisficing (RS), an optimization objective that seeks parameters which attain a target level of in-distribution performance while minimizing fragility, defined as the rate at which performance deteriorates as the data distribution departs from training. We develop a gradient-based algorithm, Iterative Robust Satisficing (IRS), that directly optimizes this criterion. Across a range of synthetic and real-world distribution shifts, including long-tailed image classification, group shifts induced by spurious correlations, and natural shifts in tabular regression, IRS consistently improves performance on minority and worst-case groups without sacrificing overall accuracy. Notably, IRS achieves these robustness gains with a per-step computational cost similar to standard stochastic gradient descent and requires only a single forward and backward pass per update. Together, these results suggest that minimizing fragility provides a practical and effective alternative to existing robust training methods for learning models that remain reliable under distribution shift.
APA
Ağırman, E., Saday, A. & Tekin, C.. (2026). Iterative Robust Satisficing: Minimizing Performance Degradation Under Distribution Shift. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:939-960 Available from https://proceedings.mlr.press/v306/agirman26a.html.

Related Material