Gt Dl L10
Readings
None!
Responsible AI
Refer to the pdf for more information.
- Bias
- Fairness and equity
- Calibration
- Provenance
- Validationand misinformation
- Robustnessand resilience
- Safety
- Security
- Privacy
- Transparency
- Interpretability
- Explainability
- Empowerment
- Redress
- Accountability and governance
Bias and Fairness Overview
FAT*
- Fairness
- Accountability
- Transparency
- and more!
Calibration and the Fairness impossibility Theorems
Definition of calibration, measuring it and the limitations of it.
The Fairness Impossibility Theorems - It is impossible for a classifier to achieve both equal calibration and error rates between groups, (if there is a difference in prevalence between the groups and the classifier is not perfect)
This post is licensed under CC BY 4.0 by the author.