Fairness in Algorithmic Decision Making
Abstract
An overview of research on fairness in algorithmic decision-making carried out at the University of Belgrade: how unfairness is measured, where it enters the modelling pipeline, and how it can be mitigated at the preprocessing, in-processing, and post-processing stages. The talk connects this work to the wider Mechanism Design for Social Good agenda of placing the interests of marginalised groups at the foundation of algorithmic and resource allocation systems.
I was invited to speak to the Mechanism Design for Social Good (MD4SG) community — now the Equity and Access in Algorithms, Mechanisms, and Optimization initiative — about our work on fairness in algorithmic decision-making.
The talk covered how unfairness is measured, where it enters a modelling pipeline, and the mitigation approaches we have developed at the preprocessing, in-processing, and post-processing stages, and how this connects to MD4SG’s broader agenda of building algorithmic and resource allocation systems around the interests of marginalised groups.
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.