Certifying and Removing Disparate Impact

2015/08/07 by Michael Feldman, Sorelle A. Friedler, John Moeller +2 · 43 citations
Business, Management and Accounting · Computer Science · Social Sciences · #Ethics and Social Impacts of AI #Intellectual Property and Patents #Law, AI, and Intellectual Property

paper · doi:10.1145/2783258.2783311

openalex publication_date 2015/08/07 · openalex created_date 2025/10/10 · openalex updated_date 2026/08/04

Abstract

What does it mean for an algorithm to be biased? In U.S. law, unintentional bias is encoded via disparate impact, which occurs when a selection process has widely different outcomes for different groups, even as it appears to be neutral. This legal determination hinges on a definition of a protected class (ethnicity, gender) and an explicit description of the process.

Citations

• Fairness through awareness

Cited by

• FACT: A Diagnostic for Group Fairness Trade-offs
• Does Object Recognition Work for Everyone?
• Learning Fairness-aware Relational Structures
• A Survey on Bias and Fairness in Machine Learning
• A Clarification of the Nuances in the Fairness Metrics Landscape
• Inclusive GAN: Improving Data and Minority Coverage in Generative Models
• No-Free-Fairness: Fundamental Limits and Trade-offs in Learning Systems
• Towards fairer datasets
• Bayes-Optimal Fair Classification with Multiple Sensitive Features
• Probabilistic Verification of Fairness Properties via Concentration
• Fair Data Representation for Machine Learning at the Pareto Frontier
• Learning Heterogeneous Performance-Fairness Trade-offs in Federated Learning
• Decision-centric fairness: Evaluation and optimization for resource allocation problems
• Semantics derived automatically from language corpora contain human-like biases
• A Design Framework for operationalizing Trustworthy Artificial Intelligence in Healthcare: Requirements, Tradeoffs and Challenges for its Clinical Adoption
• ReLU integral probability metric and its applications
• Debiasing Credit Scoring using Evolutionary Algorithms
• Algorithmic decision-making in AVs: Understanding ethical and technical concerns for smart cities
• The Bouncer Problem: Challenges to Remote Explainability
• A Framework for Understanding Sources of Harm throughout the Machine Learning Life Cycle
• Learning Fair Rule Lists
• Debiasing representations by removing unwanted variation due to protected attributes
• Whence Is A Model Fair? Fixing Fairness Bugs via Propensity Score Matching
• Feature Attributions and Counterfactual Explanations Can Be Manipulated
• Towards a critical race methodology in algorithmic fairness
• Algorithmic Fairness: Choices, Assumptions, and Definitions
• Rank-Based Multi-task Learning for Fair Regression
• Auditing Black-box Models for Indirect Influence
• BeFair: Addressing Fairness in the Banking Sector
• A Short-term Intervention for Long-term Fairness in the Labor Market
• Fairness and Accountability Design Needs for Algorithmic Support in High-Stakes Public Sector Decision-Making
• FlipTest
• Image Cropping on Twitter: Fairness Metrics, their Limitations, and the Importance of Representation, Design, and Agency
• Priority-based Post-Processing Bias Mitigation for Individual and Group Fairness
• Bias and Discrimination in AI: A Cross-Disciplinary Perspective
• Achieving Fairness in Determining Medicaid Eligibility through Fairgroup Construction
• Towards Fair Classifiers Without Sensitive Attributes
• Fairness for Robust Log Loss Classification
• Like Trainer, Like Bot? Inheritance of Bias in Algorithmic Content Moderation
• Software Engineering Principles for Fairer Systems: Experiments with GroupCART
• Towards Human-Centered Early Prediction Models for Academic Performance in Real-World Contexts
• Equity of Attention
• Sorelle Friedler [wikipedia]

Related

• Certifying and removing disparate impact
• Discrimination in the Age of Algorithms
• Discrimination in machine learning algorithms
• Fairness Under Unawareness: Assessing Disparity When Protected Class Is Unobserved
• Fairness-aware machine learning: a perspective