Scoring Functions

Cynthia Dwork
Harvard University

Scoring functions assign numbers to individuals that are frequently interpreted as probabilities, as in, “What is the probability that X will repay the loan?” However, it is not clear what is the meaning of a probability for a non-repeatable event. This has been studied for at least 50 years. Without a satisfactory answer, how can we even specify the goal of an ideal scoring algorithm? We will discuss a complexity-theoretic approach and examine it in several settings.

Back to Graduate Summer School on Algorithmic Fairness