Demystifying the Optimal Fair Classifier in Multi-Class Classification

Li Zhang, Yuyuan Li, Xiaohua Feng, Jiaming Zhang, Fengyuan Yu, Chaochao Chen
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:156364-156405, 2026.

Abstract

Ensuring fair and equitable treatment across diverse groups, particularly in multi-class classification tasks, poses a significant challenge due to the persistent biases inherent in machine learning models. Most existing bias mitigation techniques are tailored to binary settings, and the presence of multi-dimensional outputs and complex fairness mechanisms makes their extension to multi-class scenarios neither straightforward nor effective. In this paper, we investigate two fundamental, unresolved challenges in fair classification: (i) characterizing the optimal accuracy-fairness frontier in multi-class settings, and (ii) designing practical algorithms that attain this optimum across different training phases. To tackle these challenges, we first specify an analytically tractable probabilistic formulation of the optimal classifier under fairness constraints. Building upon this, we propose two attribute-blind algorithms to enforce fairness requirements in practice: an in-processing approach for fairness intervention during training via the reduction approach, and a post-processing approach for fine-tuning output probabilities with plug-in estimation. Theoretical analysis reveals that both methods converge to the optimal accuracy-fairness Pareto frontier. Experiments conducted on multiple datasets demonstrate the superior performance of our methods in balancing accuracy and fairness.

Cite this Paper


BibTeX
@InProceedings{pmlr-v306-zhang26cj, title = {Demystifying the Optimal Fair Classifier in Multi-Class Classification}, author = {Zhang, Li and Li, Yuyuan and Feng, Xiaohua and Zhang, Jiaming and Yu, Fengyuan and Chen, Chaochao}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {156364--156405}, 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/zhang26cj/zhang26cj.pdf}, url = {https://proceedings.mlr.press/v306/zhang26cj.html}, abstract = {Ensuring fair and equitable treatment across diverse groups, particularly in multi-class classification tasks, poses a significant challenge due to the persistent biases inherent in machine learning models. Most existing bias mitigation techniques are tailored to binary settings, and the presence of multi-dimensional outputs and complex fairness mechanisms makes their extension to multi-class scenarios neither straightforward nor effective. In this paper, we investigate two fundamental, unresolved challenges in fair classification: (i) characterizing the optimal accuracy-fairness frontier in multi-class settings, and (ii) designing practical algorithms that attain this optimum across different training phases. To tackle these challenges, we first specify an analytically tractable probabilistic formulation of the optimal classifier under fairness constraints. Building upon this, we propose two attribute-blind algorithms to enforce fairness requirements in practice: an in-processing approach for fairness intervention during training via the reduction approach, and a post-processing approach for fine-tuning output probabilities with plug-in estimation. Theoretical analysis reveals that both methods converge to the optimal accuracy-fairness Pareto frontier. Experiments conducted on multiple datasets demonstrate the superior performance of our methods in balancing accuracy and fairness.} }
Endnote
%0 Conference Paper %T Demystifying the Optimal Fair Classifier in Multi-Class Classification %A Li Zhang %A Yuyuan Li %A Xiaohua Feng %A Jiaming Zhang %A Fengyuan Yu %A Chaochao Chen %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-zhang26cj %I PMLR %P 156364--156405 %U https://proceedings.mlr.press/v306/zhang26cj.html %V 306 %X Ensuring fair and equitable treatment across diverse groups, particularly in multi-class classification tasks, poses a significant challenge due to the persistent biases inherent in machine learning models. Most existing bias mitigation techniques are tailored to binary settings, and the presence of multi-dimensional outputs and complex fairness mechanisms makes their extension to multi-class scenarios neither straightforward nor effective. In this paper, we investigate two fundamental, unresolved challenges in fair classification: (i) characterizing the optimal accuracy-fairness frontier in multi-class settings, and (ii) designing practical algorithms that attain this optimum across different training phases. To tackle these challenges, we first specify an analytically tractable probabilistic formulation of the optimal classifier under fairness constraints. Building upon this, we propose two attribute-blind algorithms to enforce fairness requirements in practice: an in-processing approach for fairness intervention during training via the reduction approach, and a post-processing approach for fine-tuning output probabilities with plug-in estimation. Theoretical analysis reveals that both methods converge to the optimal accuracy-fairness Pareto frontier. Experiments conducted on multiple datasets demonstrate the superior performance of our methods in balancing accuracy and fairness.
APA
Zhang, L., Li, Y., Feng, X., Zhang, J., Yu, F. & Chen, C.. (2026). Demystifying the Optimal Fair Classifier in Multi-Class Classification. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:156364-156405 Available from https://proceedings.mlr.press/v306/zhang26cj.html.

Related Material