Fairness Behind a Veil of Ignorance: A Welfare Analysis for Automated Decision Making

By Hoda Heidari, Claudio Ferrari, Krishna P. Gummadi, Andreas Krause. 2018

147 citations recorded by the source corpus.

We draw attention to an important, yet largely overlooked aspect of evaluating fairness for automated decision making systems---namely risk and welfare considerations. Our proposed family of measures corresponds to the long-established formulations of cardinal social welfare in economics, and is justified by the Rawlsian conception of fairness behind a veil of ignorance. The convex formulation of our welfare-based measures of fairness allows us to integrate them as a constraint into any convex loss minimization pipeline. Our empirical analysis reveals interesting trade-offs between our proposal and (a) prediction accuracy, (b) group discrimination, and (c) Dwork et al.'s notion of individual fairness. Furthermore and perhaps most importantly, our work provides both heuristic justification and empirical evidence suggesting that a lower-bound on our measures often leads to bounded inequality in algorithmic outcomes; hence presenting the first computationally feasible mechanism for bounding individual-level inequality.

Subfields: General Decision Sciences, Safety Research

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In General Decision Sciences

In Safety Research

Citations in the corpus

Social Data: Biases, Methodological Pitfalls, and Ethical Boundaries (2019, background): “Researchers have also developed a growing body of evaluation metrics for measuring fairness (Dwork et al., 2012; Kearns et al., 2017; Heidari et al., 2018; Verma and Rubin, 2018).”
Optimization with Non-Differentiable Constraints with Applications to Fairness, Recall, Churn, and Other Goals (2018, background): “For example, Heidari et al. (2018) give a new individual fairness notion which ensures the expected utility an individual receives as a result of the model.”
Fairness in Machine Learning: A Survey (2020, unknown): “[158, 111, 142, 132, 51, 57, 195, 290, 26, 7, 96]).”
Avoiding Resentment Via Monotonic Fairness (2019, background): “Our work also complements a body of work which explores definitions of fairness in which groups are collectively satisfied [43, 22], with variations on being a priori ambivalent or being a posteriori free of desire to switch labels as a group.”
Noise-tolerant fair classification (2019, background): “, 2019), and on means of achieving them (Zemel et al., 2013; Zafar et al., 2017b; Calmon et al., 2017; Dwork et al., 2018; Agarwal et al., 2018; Donini et al., 2018; Heidari et al., 2018).”
Fairness in Machine Learning (2020, methodology): “…used for evaluating and designing ML systems [2, 7, 14, 15, 21, 22, 23, 25, 28, 29, 31, 34, 44, 45, 46, 52, 53, 54, 55, 56, 59, 66, 69, 71, 72, 73, 74, 75, 82, 84, 87, 89, 90, 94, 95, 97, 98, 99, 103, 104, 106, 114, 115, 116, 121, 123, 126, 138, 139, 141, 148, 149, 153, 156, 159, 165, 174, 176,…”
Fair Generative Modeling via Weak Supervision (2019, methodology): “demographic parity, equality of odds and opportunity) and study methods by which models can perform tasks such as classification in a non-discriminatory way (Barocas et al., 2018; Dwork et al., 2012; Heidari et al., 2018; du Pin Calmon et al., 2018).”
Fairness risk measures (2019, background): “…constrains the subgroup error rates to be identical: There are other extant notions of perfect fairness [Zafar et al., 2017a, Ritov et al., 2017, Heidari et al., 2018, Zhang and Bareinboim, 2018], including those for individual rather than group fairness [Dwork et al., 2012, Kusner et al.,…”