Penalizing Unfairness in Binary Classification

Abstract

We present a new approach for mitigating unfairness in learned classifiers. In particular, we focus on binary classification tasks over individuals from two populations, where, as our criterion for fairness, we wish to achieve similar false positive rates in both populations, and similar false negative rates in both populations. As a proof of concept, we implement our approach and empirically evaluate its ability to achieve both fairness and accuracy, using datasets from the fields of criminal risk assessment, credit, lending, and college admissions.

Cited by in corpus (18)

• Fairness in Machine Learning: A Survey
• Fairness in Machine Learning
• Empirical Risk Minimization under Fairness Constraints
• Fair Regression for Health Care Spending
• Fairness for Robust Log Loss Classification
• A survey on datasets for fairness-aware machine learning
• Socially Responsible AI Algorithms: Issues, Purposes, and Challenges
• Learning the Pareto Front with Hypernetworks
• Putting Fairness Principles into Practice: Challenges, Metrics, and Improvements
• Review of Mathematical frameworks for Fairness in Machine Learning
• A Distributionally Robust Approach to Fair Classification
• Welfare and Distributional Impacts of Fair Classification
• Rényi Fair Inference
• Fair Classification via Domain Adaptation: A Dual Adversarial Learning Approach
• Joint Fairness Model with Applications to Risk Predictions for Under-represented Populations
• On the Impact of Data Quality on Image Classification Fairness
• Fairness-aware Summarization for Justified Decision-Making
• Optimal Pre-Processing to Achieve Fairness and Its Relationship with Total Variation Barycenter