The Fairness-Accuracy Pareto Front
arXiv:2008.10797
Abstract
Algorithmic fairness seeks to identify and correct sources of bias in machine learning algorithms. Confoundingly, ensuring fairness often comes at the cost of accuracy. We provide formal tools in this work for reconciling this fundamental tension in algorithm fairness. Specifically, we put to use the concept of Pareto optimality from multi-objective optimization and seek the fairness-accuracy Pareto front of a neural network classifier. We demonstrate that many existing algorithmic fairness methods are performing the so-called linear scalarization scheme which has severe limitations in recovering Pareto optimal solutions. We instead apply the Chebyshev scalarization scheme which is provably superior theoretically and no more computationally burdensome at recovering Pareto optimal solutions compared to the linear scheme.
arXiv commentadded toy figs to illustrate pareto optimality, some re-organization for clarity following reviewer comments
References in corpus (13)
- On Calibration of Modern Neural Networks
- Counterfactual Fairness
- Multi-Task Learning as Multi-Objective Optimization
- Data Decisions and Theoretical Implications when Adversarially Learning Fair Representations
- A Convex Framework for Fair Regression
- Achieving Fairness through Adversarial Learning: an Application to Recidivism Prediction
- From Parity to Preference-based Notions of Fairness in Classification
- Learning to Pivot with Adversarial Networks
- Fairness in Algorithmic Decision Making: An Excursion Through the Lens of Causality
- Causal Reasoning for Algorithmic Fairness
- Pareto Multi-Task Learning
- Pareto-Path Multi-Task Multiple Kernel Learning
- Random Hypervolume Scalarizations for Provable Multi-Objective Black Box Optimization