Group Fairness and Individual Fairness

Guy Rothblum
Apple Inc.

The early literature on the theory of algorithmic fairness identified two categories of fairness notions: group fairness, which requires that certain statistics be similar on different demographic groups, and individual fairness, which requires that individuals who are “similar,” with respect to a given scoring or classification task, be treated similarly. We will discuss these notions and their pros and cons.

Back to Graduate Summer School on Algorithmic Fairness