2018/07/02 by Amanda Bower, Bower, Amanda, Laura Niss +5 · 4 citations
Computer Science · Mathematics · Psychology · #Advanced Causal Inference Techniques #Advanced Statistical Methods and Models #Applied mathematics #Artificial intelligence #Computer science #Computers and Society (cs.CY) #Criminology #Debiasing #Econometrics #Explainable Artificial Intelligence (XAI) #FOS: Computer and information sciences #Machine Learning and Algorithms #Machine Learning and Data Classification #Mathematics #Orders of approximation #Psychology #Recidivism #Regression #Social psychology #Statistical Methods and Inference #Statistics #Variation (astronomy) #cs.CY
paper · pdf · doi:10.48550/arxiv.1807.00461
published in ArXiv.org (Cornell University) · Presented as a poster at the 2018 Workshop on Fairness, Accountability, and Transparency in Machine Learning (FAT/ML 2018)
arxiv created 2018/07/02 · openalex publication_date 2018/07/02 · arxiv updated 2018/07/03 · openalex created_date 2022/08/04 · openalex updated_date 2026/08/05
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.