Mitigating Bias in Models with SHAP and Fairlearn
Talkat Open AI + Data Forum @ Open Source Summit North America 2022 · on Wednesday, 2022-06-22· at 13:45· Room 408/409 (Level 4), JW Marriott Austin, 110 E 2nd Street Austin, TX 78701·
summary
Explains how to detect and reduce bias in machine learning models using SHAP for model explanations and Fairlearn for fairness assessment and mitigation. Covers concrete workflows integrating SHAP to surface feature contributions and Fairlearn to evaluate disparities and apply mitigation algorithms on models built with common tools (e.g., scikit-learn/ML frameworks).
AI-written summary · about
chapters (7)
- 0:00Start
- 2:37Reasoning about Bias
- 8:21Where to Attack Bias?
- 11:16How to Attack Bias in the Model?
- 16:25COMPAS and ProPublica Study
- 19:16COMPAS Dataset
- 31:41ThresholdOptimizer Illustration
related talks
41mMachine Learning Model Fairness and Interpretability with AWS
27mInterpretability and explainability in machine learning
36mSHAP beyond the standard graphics: co-design of ML-models in earth sciences
31mBeyond the Black Box: Interpreting ML Models with SHAP
42mFairness in Machine Learning: From Theory to Practice
47mFight AI Bias With… Bias
44mImprove explainability of ML models to meet regulatory requirements
50mEthical Machine Learning