CAFP: Fairness via Counterfactual Averaging
💡Model-agnostic fairness without retraining—theoretical guarantees for production ML.
⚡ 30-Second TL;DR
What Changed
Generates counterfactuals by flipping sensitive attributes
Why It Matters
CAFP enables fairness in deployed models without architectural changes, ideal for sensitive domains like healthcare and justice. It lowers barriers for practitioners lacking training data access.
What To Do Next
Test CAFP on your classifier by flipping sensitive attributes in test data and averaging predictions.
Key Points
- •Generates counterfactuals by flipping sensitive attributes
- •Averages factual and counterfactual predictions for fairness
- •Eliminates direct dependence on protected attributes theoretically
- •Achieves perfect demographic parity under mild assumptions
- •Reduces equalized odds gap by at least half the counterfactual bias
Weekly AI Recap
Read this week's curated digest of top AI events →
👉Related Updates
AI-curated news aggregator. All content rights belong to original publishers.
Original source: ArXiv AI ↗
This is a summary, not the original. Read the source, or get the weekly briefing.
The weekly digest
One email a week. Unsubscribe anytime.