MESD: Detecting and Mitigating Procedural Bias in Intersectional Groups

Researchers propose MESD (Multi-category Explanation Stability Disparity), a new metric to detect procedural bias in AI models across intersectional groups. They also introduce UEF framework that balances utility, explanation quality, and fairness in machine learning systems.

• →MESD addresses gaps in current bias detection by focusing on procedural fairness rather than just outcome-based metrics.
• →The metric evaluates explanation quality disparities across multiple protected categories simultaneously.
• →UEF framework uses multi-objective optimization to balance model utility, explainability, and fairness.
• →Experimental results demonstrate UEF's effectiveness in balancing competing objectives across datasets.
• →The research extends bias detection beyond single protected categories to intersectional subgroups.