Operationalizing Individual Fairness with Pairwise Fair Representations

Abstract

We revisit the notion of individual fairness proposed by Dwork et al. A central challenge in operationalizing their approach is the difficulty in eliciting a human specification of a similarity metric. In this paper, we propose an operationalization of individual fairness that does not rely on a human specification of a distance metric. Instead, we propose novel approaches to elicit and leverage side-information on equally deserving individuals to counter subordination between social groups. We model this knowledge as a fairness graph, and learn a unified Pairwise Fair Representation (PFR) of the data that captures both data-driven similarity between individuals and the pairwise side-information in fairness graph. We elicit fairness judgments from a variety of sources, including human judgments for two real-world datasets on recidivism prediction (COMPAS) and violent neighborhood prediction (Crime & Communities). Our experiments show that the PFR model for operationalizing individual fairness is practically viable.

To be published in the proceedings of the VLDB Endowment, Vol. 13, Issue. 4

References in corpus (11)

• Equality of Opportunity in Supervised Learning
• Counterfactual Fairness
• FA*IR: A Fair Top-k Ranking Algorithm
• Fairness Constraints: Mechanisms for Fair Classification
• Equity of Attention: Amortizing Individual Fairness in Rankings
• Representation Learning on Graphs: Methods and Applications
• On the (im)possibility of fairness
• A Unified Approach to Quantifying Algorithmic Unfairness: Measuring Individual & Group Unfairness via Inequality Indices
• Designing Fair Ranking Schemes
• From Parity to Preference-based Notions of Fairness in Classification
• Online Learning with an Unknown Fairness Metric

Cited by in corpus (22)

• EDITS: Modeling and Mitigating Data Bias for Graph Neural Networks
• Interplay between Upsampling and Regularization for Provider Fairness in Recommender Systems
• Equality of Learning Opportunity via Individual Fairness in Personalized Recommendations
• Fairness without Demographics through Adversarially Reweighted Learning
• A survey on datasets for fairness-aware machine learning
• Towards Fair Classifiers Without Sensitive Attributes: Exploring Biases in Related Features
• Learning Certified Individually Fair Representations
• An Empirical Study on Learning Fairness Metrics for COMPAS Data with Human Supervision
• Accounting for Model Uncertainty in Algorithmic Discrimination
• Loss-Aversively Fair Classification
• Distributional Individual Fairness in Clustering
• Removing biased data to improve fairness and accuracy
• Predictive Uncertainty-based Bias Mitigation in Ranking
• Gerrymandering Individual Fairness
• FairMILE: Towards an Efficient Framework for Fair Graph Representation Learning
• Can Querying for Bias Leak Protected Attributes? Achieving Privacy With Smooth Sensitivity
• Fairness in Ranking under Uncertainty
• Enhancing Model Robustness and Fairness with Causality: A Regularization Approach
• Beyond traditional assumptions in fair machine learning
• Correct-By-Construction: Certified Individual Fairness through Neural Network Training
• A Sociotechnical View of Algorithmic Fairness
• Individually Fair Gradient Boosting