Formalizing (and achieving?) Fairness in Machine Learning
Machine Learning techniques are a fundamental tool for automated decision systems and recommenders that substitute or support experts in a high number of decisions and fields (e.g., ranging from automated resume screening to credit score systems to criminal justice support systems).
In such a context, an increasing number of scientific studies and journalistic investigations has shown that such data-driven decision systems may have discriminating behaviors and amplify inequalities in society. In this talk we provide an overview of the problem, and we present preliminary approaches for measuring and possibly achieving fairness in ML-driven decision systems.
Prerequisites: Knowledge of R or Python, basics of probability.
#fair ML #data bias #decision systems #algorithmic discrimination