Individually Fair Ranking

2021/03/19 by Amanda Bower, Bower, Amanda, Hamid Eftekhari +5
Computer Science · Decision Sciences · Social Sciences · #Auction Theory and Applications #Domain Adaptation and Few-Shot Learning #Ethics and Social Impacts of AI #FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning (stat.ML)

paper · pdf · doi:10.48550/arxiv.2103.11023

openalex publication_date 2021/03/19 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

We develop an algorithm to train individually fair learning-to-rank (LTR) models. The proposed approach ensures items from minority groups appear alongside similar items from majority groups. This notion of fair ranking is based on the definition of individual fairness from supervised learning and is more nuanced than prior fair LTR approaches that simply ensure the ranking model provides underrepresented items with a basic level of exposure. The crux of our method is an optimal transport-based regularizer that enforces individual fairness and an efficient algorithm for optimizing the regularizer. We show that our approach leads to certifiably individually fair LTR models and demonstrate the efficacy of our method on ranking tasks subject to demographic biases.

Citations

• Equality of Opportunity in Supervised Learning
• Fairness in Criminal Justice Risk Assessments: The State of the Art
• On conditional parity as a notion of non-discrimination in machine learning
• Counterfactual Fairness
• Minimax Statistical Learning with Wasserstein Distances

Related

• Learning Fair Ranking Policies via Differentiable Optimization of Ordered Weighted Averages
• Optimizing Group-Fair Plackett-Luce Ranking Models for Relevance and Ex-Post Fairness
• Fairness Through Regularization for Learning to Rank
• Policy Learning for Fairness in Ranking
• iFair: Learning Individually Fair Data Representations for Algorithmic Decision Making