Equalized odds postprocessing under imperfect group information

Pranjal Awasthi, Matthäus Kleindessner, Jamie Morgenstern
Proceedings of the Twenty Third International Conference on Artificial Intelligence and Statistics, PMLR 108:1770-1780, 2020.

Abstract

Most approaches aiming to ensure a model’s fairness with respect to a protected attribute (such as gender or race) assume to know the true value of the attribute for every data point. In this paper, we ask to what extent fairness interventions can be effective even when only imperfect information about the protected attribute is available. In particular, we study the prominent equalized odds postprocessing method of Hardt et al. (2016) under a perturbation of the attribute. We identify conditions on the perturbation that guarantee that the bias of a classifier is reduced even by running equalized odds with the perturbed attribute. We also study the error of the resulting classifier. We empirically observe that under our identified conditions most often the error does not suffer from a perturbation of the protected attribute. For a special case, we formally prove this observation to be true.

Cite this Paper

BibTeX
@InProceedings{pmlr-v108-awasthi20a, title = {Equalized odds postprocessing under imperfect group information}, author = {Awasthi, Pranjal and Kleindessner, Matth\"aus and Morgenstern, Jamie}, booktitle = {Proceedings of the Twenty Third International Conference on Artificial Intelligence and Statistics}, pages = {1770--1780}, year = {2020}, editor = {Chiappa, Silvia and Calandra, Roberto}, volume = {108}, series = {Proceedings of Machine Learning Research}, month = {26--28 Aug}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v108/awasthi20a/awasthi20a.pdf}, url = {https://proceedings.mlr.press/v108/awasthi20a.html}, abstract = {Most approaches aiming to ensure a model’s fairness with respect to a protected attribute (such as gender or race) assume to know the true value of the attribute for every data point. In this paper, we ask to what extent fairness interventions can be effective even when only imperfect information about the protected attribute is available. In particular, we study the prominent equalized odds postprocessing method of Hardt et al. (2016) under a perturbation of the attribute. We identify conditions on the perturbation that guarantee that the bias of a classifier is reduced even by running equalized odds with the perturbed attribute. We also study the error of the resulting classifier. We empirically observe that under our identified conditions most often the error does not suffer from a perturbation of the protected attribute. For a special case, we formally prove this observation to be true.} }
Endnote
%0 Conference Paper %T Equalized odds postprocessing under imperfect group information %A Pranjal Awasthi %A Matthäus Kleindessner %A Jamie Morgenstern %B Proceedings of the Twenty Third International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2020 %E Silvia Chiappa %E Roberto Calandra %F pmlr-v108-awasthi20a %I PMLR %P 1770--1780 %U https://proceedings.mlr.press/v108/awasthi20a.html %V 108 %X Most approaches aiming to ensure a model’s fairness with respect to a protected attribute (such as gender or race) assume to know the true value of the attribute for every data point. In this paper, we ask to what extent fairness interventions can be effective even when only imperfect information about the protected attribute is available. In particular, we study the prominent equalized odds postprocessing method of Hardt et al. (2016) under a perturbation of the attribute. We identify conditions on the perturbation that guarantee that the bias of a classifier is reduced even by running equalized odds with the perturbed attribute. We also study the error of the resulting classifier. We empirically observe that under our identified conditions most often the error does not suffer from a perturbation of the protected attribute. For a special case, we formally prove this observation to be true.
APA
Awasthi, P., Kleindessner, M. & Morgenstern, J.. (2020). Equalized odds postprocessing under imperfect group information. Proceedings of the Twenty Third International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 108:1770-1780 Available from https://proceedings.mlr.press/v108/awasthi20a.html.

Related Material