Training Individually Fair ML Models with Sensitive Subspace Robustness
Abstract
We consider training machine learning models that are fair in the sense that their performance is invariant under certain sensitive perturbations to the inputs. For example, the performance of a resume screening system should be invariant under changes to the gender and/or ethnicity of the applicant. We formalize this notion of algorithmic fairness as a variant of individual fairness and develop a distributionally robust optimization approach to enforce it during training. We also demonstrate the effectiveness of the approach on two ML tasks that are susceptible to gender and racial biases.
Cite
Text
Yurochkin et al. "Training Individually Fair ML Models with Sensitive Subspace Robustness." International Conference on Learning Representations, 2020.Markdown
[Yurochkin et al. "Training Individually Fair ML Models with Sensitive Subspace Robustness." International Conference on Learning Representations, 2020.](https://mlanthology.org/iclr/2020/yurochkin2020iclr-training/)BibTeX
@inproceedings{yurochkin2020iclr-training,
title = {{Training Individually Fair ML Models with Sensitive Subspace Robustness}},
author = {Yurochkin, Mikhail and Bower, Amanda and Sun, Yuekai},
booktitle = {International Conference on Learning Representations},
year = {2020},
url = {https://mlanthology.org/iclr/2020/yurochkin2020iclr-training/}
}