Constraining deep representations with a noise module for fair classification

Abstract

The recent surge in interest for Deep Learning (motivated by its exceptional performances on longstanding problems) made Neural Networks a very appealing tool for many actors in our society. One issue in this shift of interest is that Neural Networks are very opaque objects and it is often hard to make sense of their predictions. In this context, research efforts have focused on building fair representations of data which display little to no correlation with regard to a sensitive attribute s. In this paper we build onto a domain adaptation neural model by augmenting it with a "noise conditioning" mechanism which we show is instrumental in obtaining fair (i.e. non-correlated with s) representations. We provide experiments on standard datasets showing the effectiveness of the noise conditioning mechanism in helping the networks to ignore the sensible attribute.