Within the last few years, researchers have come to understand that machine learning systems may display discriminatory behavior with regards to certain protected characteristics, such as gender or race. As a result, many definitions of fairness have been created to enable equity in machine learning algorithms, each with their own strengths and weaknesses.
1.) Overview of Fairness, Accountability, and Transparency (FAT*) in Machine Learning. (5 minutes) 2.) Measures of Algorithmic Fairness (10 minutes) a.) Anti-Classification Measures b.) Parity Measures c.) Calibration Measures 3.) Strengths and Weakness of different types of fairness measures. (5 minutes) 4.) Measuring the Delayed Impact of Fair Machine Learning. (12 minutes) 5.) Theoretical Limitations of Algorithmic Fairness. (7 minutes) 6.) Conclusion/Questions (2 minutes)