Fairness Evaluation in Presence of Biased Noisy Labels

Abstract

Risk assessment tools are widely used around the country to inform decision making within the criminal justice system. Recently, considerable attention has been devoted to the question of whether such tools may suffer from racial bias. In this type of assessment, a fundamental issue is that the training and evaluation of the model is based on a variable (arrest) that may represent a noisy version of an unobserved outcome of more central interest (offense). We propose a sensitivity analysis framework for assessing how assumptions on the noise across groups affect the predictive bias properties of the risk assessment model as a predictor of reoffense. Our experimental results on two real world criminal justice data sets demonstrate how even small biases in the observed labels may call into question the conclusions of an analysis based on the noisy outcome.

arXiv commentAccepted at International Conference on Artificial Intelligence and Statistics (AISTATS), 2020

References in corpus (10)

• Equality of Opportunity in Supervised Learning
• Counterfactual Fairness
• Classification with Noisy Labels by Importance Reweighting
• The Measure and Mismeasure of Fairness
• Residual Unfairness in Fair Machine Learning from Prejudiced Data
• Causal Reasoning for Algorithmic Fairness
• Instance-Dependent PU Learning by Bayesian Optimal Relabeling
• Learning from Binary Labels with Instance-Dependent Corruption
• Recovering True Classifier Performance in Positive-Unlabeled Learning
• A Generalized Neyman-Pearson Criterion for Optimal Domain Adaptation

Cited by in corpus (6)

• It's COMPASlicated: The Messy Relationship between RAI Datasets and Algorithmic Fairness Benchmarks
• Uncertainty as a Form of Transparency: Measuring, Communicating, and Using Uncertainty
• Fairness On The Ground: Applying Algorithmic Fairness Approaches to Production Systems
• Hierarchical Bayesian Models to Mitigate Systematic Disparities in Prediction with Proxy Outcomes
• Can Less be More? When Increasing-to-Balancing Label Noise Rates Considered Beneficial
• Measuring Fairness Under Unawareness of Sensitive Attributes: A Quantification-Based Approach