Fairness in Machine Learning – A Hands-On Tutorial
Friday, September 20, 2019 — 10:40AM - 11:40AM
Machine learning (ML) drives an increasing number of economic, scientific, and social decisions. Therefore, a practical understanding of ML is crucial for students in computer science and related disciplines. Unfortunately, much work has shown that black-box ML models reproduce the biases found in data, e.g., by race, gender, or other protected attributes. This tutorial gives students without previous ML experience a hands-on understanding of how detrimental biases on race, gender or other data affects real-world ML applications. We introduce fair ML models which counter-act these biases. Students should come away with (1) an understanding of how underlying data introduces detrimental bias, (2) a hands-on understanding of the ML models used and (3) high-level challenges for incorporating fairness in ML.