Repairing without Retraining: Avoiding Disparate Impact with Counterfactual Distributions

Abstract

When the performance of a machine learning model varies over groups defined by sensitive attributes (e.g., gender or ethnicity), the performance disparity can be expressed in terms of the probability distributions of the input and output variables over each group. In this paper, we exploit this fact to reduce the disparate impact of a fixed classification model over a population of interest. Given a black-box classifier, we aim to eliminate the performance gap by perturbing the distribution of input variables for the disadvantaged group. We refer to the perturbed distribution as a counterfactual distribution, and characterize its properties for common fairness criteria. We introduce a descent algorithm to learn a counterfactual distribution from data. We then discuss how the estimated distribution can be used to build a data preprocessor that can reduce disparate impact without training a new model. We validate our approach through experiments on real-world datasets, showing that it can repair different forms of disparity without a significant drop in accuracy.

Cited by in corpus (12)

• Fairness in Machine Learning: A Survey
• Towards Out-Of-Distribution Generalization: A Survey
• EDITS: Modeling and Mitigating Data Bias for Graph Neural Networks
• Fairness via Representation Neutralization
• Algorithmic Decision Making with Conditional Fairness
• Rethinking Influence Functions of Neural Networks in the Over-parameterized Regime
• Fair Classification with Noisy Protected Attributes: A Framework with Provable Guarantees
• Information Theoretic Measures for Fairness-aware Feature Selection
• Fairness with Continuous Optimal Transport
• Multi-Stage Influence Function
• Linear Classifiers that Encourage Constructive Adaptation
• Enabling SQL-based Training Data Debugging for Federated Learning