A Game-Theoretic Framework for Measuring and Explaining Metric Compatibility in Fair Machine Learning

Lingfeng Zhang, Jingran Yang, Zhaohui Wang, Min Zhang, Zhang Qing
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:159887-159915, 2026.

Abstract

Machine learning fairness research documents trade-offs but lacks quantitative frameworks to measure intrinsic metric compatibility without requiring causal graphs. We introduce a game-theoretic framework that decomposes metrics into interaction vectors, enabling compatibility measurement between metrics via cosine similarity and mechanistic attribution to attribute coalitions. Through analysis of 6 datasets, 7 models, and 6 debiasing methods, we reveal that fairness and utility are often structurally orthogonal (median compatibility $\approx 0$) rather than diametrically opposed, with conflicts driven by sparse, low-order interactions. We further show that debiasing improves fairness by compressing the compatibility space—reducing compatibility of both synergistic and conflicting relationships—rather than eliminating conflicts, providing a mechanistic basis for understanding metric alignment.

Cite this Paper


BibTeX
@InProceedings{pmlr-v306-zhang26hp, title = {A Game-Theoretic Framework for Measuring and Explaining Metric Compatibility in Fair Machine Learning}, author = {Zhang, Lingfeng and Yang, Jingran and Wang, Zhaohui and Zhang, Min and Qing, Zhang}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {159887--159915}, 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/zhang26hp/zhang26hp.pdf}, url = {https://proceedings.mlr.press/v306/zhang26hp.html}, abstract = {Machine learning fairness research documents trade-offs but lacks quantitative frameworks to measure intrinsic metric compatibility without requiring causal graphs. We introduce a game-theoretic framework that decomposes metrics into interaction vectors, enabling compatibility measurement between metrics via cosine similarity and mechanistic attribution to attribute coalitions. Through analysis of 6 datasets, 7 models, and 6 debiasing methods, we reveal that fairness and utility are often structurally orthogonal (median compatibility $\approx 0$) rather than diametrically opposed, with conflicts driven by sparse, low-order interactions. We further show that debiasing improves fairness by compressing the compatibility space—reducing compatibility of both synergistic and conflicting relationships—rather than eliminating conflicts, providing a mechanistic basis for understanding metric alignment.} }
Endnote
%0 Conference Paper %T A Game-Theoretic Framework for Measuring and Explaining Metric Compatibility in Fair Machine Learning %A Lingfeng Zhang %A Jingran Yang %A Zhaohui Wang %A Min Zhang %A Zhang Qing %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-zhang26hp %I PMLR %P 159887--159915 %U https://proceedings.mlr.press/v306/zhang26hp.html %V 306 %X Machine learning fairness research documents trade-offs but lacks quantitative frameworks to measure intrinsic metric compatibility without requiring causal graphs. We introduce a game-theoretic framework that decomposes metrics into interaction vectors, enabling compatibility measurement between metrics via cosine similarity and mechanistic attribution to attribute coalitions. Through analysis of 6 datasets, 7 models, and 6 debiasing methods, we reveal that fairness and utility are often structurally orthogonal (median compatibility $\approx 0$) rather than diametrically opposed, with conflicts driven by sparse, low-order interactions. We further show that debiasing improves fairness by compressing the compatibility space—reducing compatibility of both synergistic and conflicting relationships—rather than eliminating conflicts, providing a mechanistic basis for understanding metric alignment.
APA
Zhang, L., Yang, J., Wang, Z., Zhang, M. & Qing, Z.. (2026). A Game-Theoretic Framework for Measuring and Explaining Metric Compatibility in Fair Machine Learning. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:159887-159915 Available from https://proceedings.mlr.press/v306/zhang26hp.html.

Related Material