Distributionally Robust Set Representation Learning Under Inference-Time Element Corruption

Yankai Chen, Hanrong Zhang, Bowei He, Philip S. Yu, Xue Liu
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:18166-18187, 2026.

Abstract

Standard Set Representation Learning methods typically excel on curated data but often overlook the challenge of Inference-time Element Corruption. This refers to scenarios where deployed models encounter element-level degradations, such as outliers or missing components, that may distort set representation and degrade performance. We propose SW-DRSO, a distributionally robust optimization framework tailored for sets. Rather than minimizing loss solely on observed training data, SW-DRSO optimizes a tractable surrogate of the worst-case expected loss over a family of plausible inference-time variations. We introduce a barycentric adversary that approximates the intractable search over corrupted sets by a differentiable training-time optimization over simplex weights. Extensive experiments across four tasks demonstrate that SW-DRSO effectively enhances robustness against corruption while maintaining high overall performance.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-chen26gg, title = {Distributionally Robust Set Representation Learning Under Inference-Time Element Corruption}, author = {Chen, Yankai and Zhang, Hanrong and He, Bowei and Yu, Philip S. and Liu, Xue}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {18166--18187}, 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/chen26gg/chen26gg.pdf}, url = {https://proceedings.mlr.press/v306/chen26gg.html}, abstract = {Standard Set Representation Learning methods typically excel on curated data but often overlook the challenge of Inference-time Element Corruption. This refers to scenarios where deployed models encounter element-level degradations, such as outliers or missing components, that may distort set representation and degrade performance. We propose SW-DRSO, a distributionally robust optimization framework tailored for sets. Rather than minimizing loss solely on observed training data, SW-DRSO optimizes a tractable surrogate of the worst-case expected loss over a family of plausible inference-time variations. We introduce a barycentric adversary that approximates the intractable search over corrupted sets by a differentiable training-time optimization over simplex weights. Extensive experiments across four tasks demonstrate that SW-DRSO effectively enhances robustness against corruption while maintaining high overall performance.} }
Endnote
%0 Conference Paper %T Distributionally Robust Set Representation Learning Under Inference-Time Element Corruption %A Yankai Chen %A Hanrong Zhang %A Bowei He %A Philip S. Yu %A Xue Liu %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-chen26gg %I PMLR %P 18166--18187 %U https://proceedings.mlr.press/v306/chen26gg.html %V 306 %X Standard Set Representation Learning methods typically excel on curated data but often overlook the challenge of Inference-time Element Corruption. This refers to scenarios where deployed models encounter element-level degradations, such as outliers or missing components, that may distort set representation and degrade performance. We propose SW-DRSO, a distributionally robust optimization framework tailored for sets. Rather than minimizing loss solely on observed training data, SW-DRSO optimizes a tractable surrogate of the worst-case expected loss over a family of plausible inference-time variations. We introduce a barycentric adversary that approximates the intractable search over corrupted sets by a differentiable training-time optimization over simplex weights. Extensive experiments across four tasks demonstrate that SW-DRSO effectively enhances robustness against corruption while maintaining high overall performance.
APA
Chen, Y., Zhang, H., He, B., Yu, P.S. & Liu, X.. (2026). Distributionally Robust Set Representation Learning Under Inference-Time Element Corruption. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:18166-18187 Available from https://proceedings.mlr.press/v306/chen26gg.html.

Related Material