Learning Fair Classifiers

By Muhammad Bilal Zafar, Isabel Valera, Manuel Gomez Rodriguez, Krishna P. Gummadi. 2015

50 citations recorded by the source corpus.

Automated data-driven decision systems are ubiquitous across a wide variety of online services, from online social networking and e-commerce to e-government. These systems rely on complex learning methods and vast amounts of data to optimize the service functionality, satisfaction of the end user and profitability. However, there is a growing concern that these automated decisions can lead to user discrimination, even in the absence of intent, leading to a lack of fairness, i.e., their outcomes have a disproportionally large adverse impact on particular groups of people sharing one or more sensitive attributes (e.g., race, sex). In this paper, we introduce a flexible mechanism to design fair classifiers in a principled manner. Then, we instantiate this mechanism on three well-known classifiers -- logistic regression, hinge loss and linear and nonlinear support vector machines. Experiments on both synthetic and real-world data show that our mechanism allows for a fine-grained control of the level of fairness, often at a minimal cost in terms of accuracy, and it provides more flexibility than alternatives.

Subfields: Safety Research, Safety, Robustness & Interpretability

How researchers see this paper

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

In Safety Research

In Safety, Robustness & Interpretability

Citations in the corpus

Concrete Problems in AI Safety (2016, background): “• Fairness: How can we make sure ML systems don’t discriminate? [3, 163, 6, 45, 117, 164]”
Does Object Recognition Work for Everyone? (2019, background): “Because statistical parity may be inappropriate when base rates differ between groups, disparate impact [15, 47] poses that positiveclassification rates between any two groups should not vary by more than 80%.”
On Fairness and Calibration (2017, background): “A related notion is Disparate Impact [14, 34], which states that the prediction rates for any two groups should not differ by more than 80% (a number motivated by legal requirements).”
Fairness Beyond Disparate Treatment & Disparate Impact: Learning Classification without Disparate Mistreatment (2016, methodology): “A number of recent studies have focused on designing decision making systems which avoid one or both of the above types of unfairness [9, 12, 13, 22].”
Fairness in Decision-Making - The Causal Explanation Formula (2018, background): “One popular criterion for the outcome fairness is the demographic parity (Zafar et al. 2015), measured by the total variation:”
Learning Controllable Fair Representations (2018, background): “For conciseness, we focus on the demographic parity notion of fairness (Calders et al., 2009; Zliobaite, 2015; Zafar et al., 2015), which requires the decisions made by a classifier over z to be independent of the sensitive attributes u.”
The Case for Process Fairness in Learning: Feature Selection for Fair Decision Making (2016, background): “These concerns have spawned a flurry of recent research activity into learning methods for detecting and avoiding unfairness in decision making [8, 9, 12, 15, 16, 18, 19].”
On preserving non-discrimination when combining expert advice (2018, background): “[20] or more directly during training from batch data [41, 19, 39, 40, 3] which requires care due to the predictors being discriminatory with respect to the particular metric of interest.”
Who Makes Trends? Understanding Demographic Biases in Crowdsourced Recommendations (2017, background): “More broadly, our work offers a new perspective to the growing debate about fairness, bias, and transparency of decisions taken by algorithms operating over big crowd-sourced data (Zafar et al. 2015; Chakraborty et al. 2016; Kroll 2015).”
Fair, Transparent, and Accountable Algorithmic Decision-making Processes (2017, background): “There is ample evidence of detrimental impacts of current non-algorithmic approaches to access to ﬁnance, employment, and housing.”
The cost of fairness in binary classification (2018, methodology): “This fairness-aware learning problem has received considerable attention in the machine learning community of late (Pedreshi et al., 2008; Kamiran and Calders, 2009; Calders and Verwer, 2010; Dwork et al., 2012; Kamishima et al., 2012; Fukuchi et al., 2013; Zafar et al., 2016; Hardt et al., 2016; Za”
Learning to Pivot with Adversarial Networks (2016, background): “…to the problem of enforcing fairness in classiﬁcation (Kamishima et al., 2012; Zemel et al., 2013; Feldman et al., 2015; Edwards and Storkey, 2015; Zafar et al., 2015; Louizos et al., 2015), which is stated as learning a classiﬁer that is independent of some chosen attribute such as gender,…”