AIAI Seminar-06 June-Talk by Mattia Cerrato (Institute of Computer Science)

Title: 10 Years of Fair Representations

Abstract: Research on algorithmic fairness has focused Fair Representation Learning (FRL) is a broad set of techniques, mostly based on neural networks, that seeks to learn new representations of data in which sensitive or undesired information has been removed. In this talk, I will offer a look back at the first ten years of FRL and its applications. Then, I will present a theoretical impossibility result for FRL with deterministic neural networks and the results of a massive experimentation we performed (around 300k model fits) supporting this result.