Algorithmic Fairness: Choices, Assumptions, and Definitions

Algorithmic Fairness: Choices, Assumptions, and Definitions

Keywords algorithmic fairness, predictive modeling, statistical learning, machine learning, decision theory

Abstract

1. INTRODUCTION

2. CHOICES, ASSUMPTIONS, AND CONSIDERATIONS

2.1. The Policy Question

2.2. The Statistical Learning Problem

2.3. Axes of Fairness and Protected Groups

3. SETUP AND NOTATION

4. FLAVORS OF FAIRNESS DEFINITIONS FROM DATA ALONE

4.1. Unconstrained Utility Maximization and Single-Threshold Fairness

4.2. Equal Prediction Measures

4.3. Equal Decision Measures

4.4. Impossibilities

5. FLAVORS OF FAIRNESS DEFINITIONS INCORPORATING ADDITIONAL CONTEXT

5.1. Metric Fairness

5.2. Causal Definitions

6. WAYS FORWARD

7. CONCLUSION

DISCLOSURE STATEMENT

ACKNOWLEDGMENTS

LITERATURE CITED

Notes