A new framework integrates fairness and explainability in machine learning by utilizing Shapley values. This approach allows for the measurement and explanation of unfairness, identifying features that contribute to it. The method has been demonstrated on the Census Income dataset, showing that 'Age', 'Number of hours', and 'Marital status' contribute to gender unfairness more efficiently than traditional Bootstrap tests. AI
IMPACT This framework could lead to more robust and interpretable AI systems by providing a unified approach to fairness and explainability.
RANK_REASON The item is a research paper detailing a new framework for measuring and explaining unfairness in machine learning. [lever_c_demoted from research: ic=1 ai=1.0]
- Age
- arXiv
- Bootstrap Tests for Distributional Treatment Effects in Instrumental Variable Models
- gender
- marital status
- Number of hours usable for astronomical observation at several sites in Mongolia
- Rawls
- Shapley value
- Stéphane Mussard
- UCI Machine Learning Repository
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →