DistMatch: Adaptive Binning via Distribution Matching for Robust Sequential Conformal Prediction

Enver Menadjiev, Jihyeon Seong, Jisu Yeo, Jaesik Choi
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:87680-87713, 2026.

Abstract

Sequential conformal prediction (CP) provides valid uncertainty quantification under the assumption of residual exchangeability. However, this assumption is often violated in real-world time series due to temporal dependencies and distributional shifts. While recent methods attempt to approximate exchangeability through reweighting, identifying optimal weights remains an open challenge. To address this limitation, we propose DistMatch, a binning-based method that recursively partitions residuals within a binary tree using the Kolmogorov–Smirnov (KS) statistic. We theoretically show that this partitioning induces approximately exchangeable leaves, thereby avoiding the need for reweighting. By applying quantile regression with online updates within each leaf, DistMatch enables locally adaptive inference and improves robustness to distributional shifts. Extensive experiments demonstrate that DistMatch outperforms existing sequential CP methods.

Cite this Paper


BibTeX
@InProceedings{pmlr-v306-menadjiev26a, title = {{D}ist{M}atch: Adaptive Binning via Distribution Matching for Robust Sequential Conformal Prediction}, author = {Menadjiev, Enver and Seong, Jihyeon and Yeo, Jisu and Choi, Jaesik}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {87680--87713}, 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/menadjiev26a/menadjiev26a.pdf}, url = {https://proceedings.mlr.press/v306/menadjiev26a.html}, abstract = {Sequential conformal prediction (CP) provides valid uncertainty quantification under the assumption of residual exchangeability. However, this assumption is often violated in real-world time series due to temporal dependencies and distributional shifts. While recent methods attempt to approximate exchangeability through reweighting, identifying optimal weights remains an open challenge. To address this limitation, we propose DistMatch, a binning-based method that recursively partitions residuals within a binary tree using the Kolmogorov–Smirnov (KS) statistic. We theoretically show that this partitioning induces approximately exchangeable leaves, thereby avoiding the need for reweighting. By applying quantile regression with online updates within each leaf, DistMatch enables locally adaptive inference and improves robustness to distributional shifts. Extensive experiments demonstrate that DistMatch outperforms existing sequential CP methods.} }
Endnote
%0 Conference Paper %T DistMatch: Adaptive Binning via Distribution Matching for Robust Sequential Conformal Prediction %A Enver Menadjiev %A Jihyeon Seong %A Jisu Yeo %A Jaesik Choi %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-menadjiev26a %I PMLR %P 87680--87713 %U https://proceedings.mlr.press/v306/menadjiev26a.html %V 306 %X Sequential conformal prediction (CP) provides valid uncertainty quantification under the assumption of residual exchangeability. However, this assumption is often violated in real-world time series due to temporal dependencies and distributional shifts. While recent methods attempt to approximate exchangeability through reweighting, identifying optimal weights remains an open challenge. To address this limitation, we propose DistMatch, a binning-based method that recursively partitions residuals within a binary tree using the Kolmogorov–Smirnov (KS) statistic. We theoretically show that this partitioning induces approximately exchangeable leaves, thereby avoiding the need for reweighting. By applying quantile regression with online updates within each leaf, DistMatch enables locally adaptive inference and improves robustness to distributional shifts. Extensive experiments demonstrate that DistMatch outperforms existing sequential CP methods.
APA
Menadjiev, E., Seong, J., Yeo, J. & Choi, J.. (2026). DistMatch: Adaptive Binning via Distribution Matching for Robust Sequential Conformal Prediction. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:87680-87713 Available from https://proceedings.mlr.press/v306/menadjiev26a.html.

Related Material