fairlearn/fairlearn
A Python package to assess and improve fairness of machine learning models.
What it solves
Fairlearn is designed to address unfairness in AI systems, specifically focusing on "group fairness." It helps developers identify and reduce harms—such as allocation harms (where opportunities or resources are withheld) and quality-of-service harms (where a system performs poorly for specific groups of people).
How it works
The package provides two primary tools:
- Metrics: Tools to assess which groups are negatively impacted by a model and compare different models using fairness and accuracy metrics.
- Algorithms: Mitigation algorithms that reduce unfairness across various AI tasks based on different definitions of fairness.
Who it’s for
It is built for AI developers and data scientists who need to quantify and mitigate bias in their machine learning models to ensure equitable outcomes across different demographic groups.
Highlights
- own a set of mitigation algorithms to reduce bias
- provides metrics for model assessment
- focuses on both allocation and quality-of-service harms
- includes Jupyter notebooks with practical usage examples
Related
- Project
- Project
- Project
- Project
- Project