Fairness and Equity in Algorithmic Decision-Making

Abstract

This chapter explores the issue of algorithmic fairness, particularly in the context of automated decision-making systems and machine learning algorithms increasingly used in areas such as employment, justice, and education. Through the analysis of well-known examples of unfair algorithms, it is demonstrated how biased data and models can lead to discriminatory decisions. The chapter also provides a comprehensive literature review on addressing algorithmic fairness, including data preprocessing, algorithm adaptation, and post-prediction processing. The work presents the key results achieved by the authors in this field. Experimental results show that there is an inevitable trade-off between fairness and accuracy regardless of the approach, but it is shown that a slight reduction in accuracy can significantly improve fairness, especially at the group level. It is emphasized that further research is needed on methods that combine mathematical and social aspects of fairness, as well as the development of algorithmic systems that enable transparency and accountability in decision-making.

Sandro Radovanović

Assistant Professor at University of Belgrade

My research interests include machine learning, development and design of decision support systems, decision theory, and fairness and justice concepts in algorithmic decision making.