www-ai.cs.tu-dortmund.de/LEHRE/SEMINARE/WS2021/TrustworthyAIMachineLearning/literature/zafar2019.pdf
outcomes (Kamiran and Calders, 2009; Corbett-Davies et al., 2017; Hardt et al., 2016; Feldman et al., 2015; Zafar et al., 2017b,a; Zemel et al., 2013; Dwork et al., 2012). In doing so, they have have measured [...] end
Furthermore, for training the classifiers, we consider a similar set of features as Goel et al. (2015). After performing these two filtering steps, we obtain 5, 832 subjects and 19 features.
Methods. [...] 2009; Kleinberg et al., 2017a), product personalization (Covington et al., 2016; Gomez-Uribe and Hunt, 2015) and online ad delivery (Google AdSense).
Several recent studies have shown that algorithmic decision …