From Parity to Preference-based Notions of Fairness in Classification

Abstract

The adoption of automated, data-driven decision making in an ever expanding range of applications has raised concerns about its potential unfairness towards certain social groups. In this context, a number of recent studies have focused on defining, detecting, and removing unfairness from data-driven decision systems. However, the existing notions of fairness, based on parity (equality) in treatment or outcomes for different social groups, tend to be quite stringent, limiting the overall decision making accuracy. In this paper, we draw inspiration from the fair-division and envy-freeness literature in economics and game theory and propose preference-based notions of fairness -- given the choice between various sets of decision treatments or outcomes, any group of users would collectively prefer its treatment or outcomes, regardless of the (dis)parity as compared to the other groups. Then, we introduce tractable proxies to design margin-based classifiers that satisfy these preference-based notions of fairness. Finally, we experiment with a variety of synthetic and real-world datasets and show that preference-based fairness allows for greater decision accuracy than parity-based fairness.

arXiv commentTo appear in Proceedings of the 31st Conference on Neural Information Processing Systems (NIPS 2017). Code available at: https://github.com/mbilalzafar/fair-classification

Cited by in corpus (45)

• Fairness in Machine Learning
• 50 Years of Test (Un)fairness: Lessons for Machine Learning
• Attack of the Tails: Yes, You Really Can Backdoor Federated Learning
• Operationalizing Individual Fairness with Pairwise Fair Representations
• Empirical Risk Minimization under Fairness Constraints
• Toward Pareto Efficient Fairness-Utility Trade-off inRecommendation through Reinforcement Learning
• Aequitas: A Bias and Fairness Audit Toolkit
• Explainable Fairness in Recommendation
• Fairness without Demographics through Adversarially Reweighted Learning
• Fairness for Robust Log Loss Classification
• Getting a CLUE: A Method for Explaining Uncertainty Estimates
• Improving Fairness via Federated Learning
• Blind Justice: Fairness with Encrypted Sensitive Attributes
• Noise-tolerant fair classification
• Classification with Fairness Constraints: A Meta-Algorithm with Provable Guarantees
• A General Framework for Fair Regression
• Fairness risk measures
• Fairness in Agreement With European Values: An Interdisciplinary Perspective on AI Regulation
• Review of Mathematical frameworks for Fairness in Machine Learning
• Toward Operationalizing Pipeline-aware ML Fairness: A Research Agenda for Developing Practical Guidelines and Tools
• Fairness Through Robustness: Investigating Robustness Disparity in Deep Learning
• Envy-Free Classification
• Accounting for Model Uncertainty in Algorithmic Discrimination
• Loss-Aversively Fair Classification
• Learning Fair Policies in Decentralized Cooperative Multi-Agent Reinforcement Learning
• Towards Fairness in Personalized Ads Using Impression Variance Aware Reinforcement Learning
• Ethical Adversaries: Towards Mitigating Unfairness with Adversarial Machine Learning
• The Fairness-Accuracy Pareto Front
• Pareto Efficient Fairness in Supervised Learning: From Extraction to Tracing
• Fair Clustering with Multiple Colors
• ProPublica's COMPAS Data Revisited
• Principal Fairness: Removing Bias via Projections
• Avoiding Resentment Via Monotonic Fairness
• Unintended Selection: Persistent Qualification Rate Disparities and Interventions
• Metrics and methods for a systematic comparison of fairness-aware machine learning algorithms
• Multi-Category Fairness in Sponsored Search Auctions
• Learning Fair and Interpretable Representations via Linear Orthogonalization
• The invisible power of fairness. How machine learning shapes democracy
• Adversarial Regression with Doubly Non-negative Weighting Matrices
• Inherent Trade-offs in the Fair Allocation of Treatments
• Fairness-aware Summarization for Justified Decision-Making
• Fairness Preferences, Actual and Hypothetical: A Study of Crowdworker Incentives
• A Sociotechnical View of Algorithmic Fairness
• On the Identification of Fair Auditors to Evaluate Recommender Systems based on a Novel Non-Comparative Fairness Notion
• Non-Comparative Fairness for Human-Auditing and Its Relation to Traditional Fairness Notions