Debiasing representations by removing unwanted variation due to protected attributes
arXiv:1807.00461
Abstract
We propose a regression-based approach to removing implicit biases in representations. On tasks where the protected attribute is observed, the method is statistically more efficient than known approaches. Further, we show that this approach leads to debiased representations that satisfy a first order approximation of conditional parity. Finally, we demonstrate the efficacy of the proposed approach by reducing racial bias in recidivism risk scores.
arXiv commentPresented as a poster at the 2018 Workshop on Fairness, Accountability, and Transparency in Machine Learning (FAT/ML 2018)
References in corpus (6)
- Equality of Opportunity in Supervised Learning
- Man is to Computer Programmer as Woman is to Homemaker? Debiasing Word Embeddings
- On the (im)possibility of fairness
- Fairness in Learning: Classic and Contextual Bandits
- A statistical framework for fair predictive algorithms
- On conditional parity as a notion of non-discrimination in machine learning