An Empirical Study of Rich Subgroup Fairness for Machine Learning

By Michael Kearns, Seth Neel, Aaron Roth, Zhiwei Steven Wu. 2018

228 citations recorded by the source corpus.

Kearns, Neel, Roth, and Wu [ICML 2018] recently proposed a notion of rich subgroup fairness intended to bridge the gap between statistical and individual notions of fairness. Rich subgroup fairness picks a statistical fairness constraint (say, equalizing false positive rates across protected groups), but then asks that this constraint hold over an exponentially or infinitely large collection of subgroups defined by a class of functions with bounded VC dimension. They give an algorithm guaranteed to learn subject to this constraint, under the condition that it has access to oracles for perfectly learning absent a fairness constraint. In this paper, we undertake an extensive empirical evaluation of the algorithm of Kearns et al. On four real datasets for which fairness is a concern, we investigate the basic convergence of the algorithm when instantiated with fast heuristics in place of learning oracles, measure the tradeoffs between fairness and accuracy, and compare this approach with the recent algorithm of Agarwal, Beygelzeimer, Dudik, Langford, and Wallach [ICML 2018], which implements weaker and more traditional marginal fairness constraints defined by individual protected attributes. We find that in general, the Kearns et al. algorithm converges quickly, large gains in fairness can be obtained with mild costs to accuracy, and that optimizing accuracy subject only to marginal fairness leads to classifiers with substantial subgroup unfairness. We also provide a number of analyses and visualizations of the dynamics and behavior of the Kearns et al. algorithm. Overall we find this algorithm to be effective on real data, and rich subgroup fairness to be a viable notion in practice.

Subfields: Artificial Intelligence, Safety Research, Evaluation & Benchmarks, AI Ethics, Fairness & Society

How researchers see this paper

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

In Artificial Intelligence

In Safety Research

In Evaluation & Benchmarks

In AI Ethics, Fairness & Society

Citations in the corpus

A Survey on Bias and Fairness in Machine Learning (2019, background): “It picks a group fairness constraint like equalizing false positive and asks that this constraint holds over an exponentially large collection of subgroups [66, 67].”
Preventing Fairness Gerrymandering: Auditing and Learning for Subgroup Fairness (2017, methodology): “Full details are contained in Section 5, and for a substantially more comprehensive empirical investigation of our method we direct the interested reader to Kearns et al. [2018].”
Flexibly Fair Representation Learning by Disentanglement (2019, positive): “(2018)—with empirical follow-up (Kearns et al., 2019)— and Hebert-Johnson et al.”
The Frontiers of Fairness in Machine Learning (2018, background): “Two recent papers, [KNRW18b] and [HJKRR18] (see also [KNRW18a, KGZ18] for empirical evaluations of the algorithms proposed in these papers), attempt to do this by asking for statistical fairness deﬁnitions to hold not just on a small number of protected groups, but on an exponential or inﬁnite…”
Learning Fair Rule Lists (2019, background): “As suggested in recent works [34, 35], providing the Pareto frontier instead of a unique model gives the possibility to the stakeholders or policymakers to select the right model in a domain-specific manner.”
Towards Intersectionality in Machine Learning: Including More Identities, Handling Underrepresentation, and Performing Evaluation (2022, methodology): “intersectional fairness loss term, GRP [121] is a probabilistic combination of models, and GRY [76, 77] produces costsensitive classifications from a 2-player zero-sum game.”
Average Individual Fairness: Algorithms, Generalization and Experiments (2019, background): “This effect isn’t merely hypothetical — Kearns et al. (2018, 2019) showed similar effects when using off-the-shelf fairness constrained learning techniques on real datasets.”
Equity Beyond Bias in Language Technologies for Education (2019, background): “Research in computational ethics (Hooker and Kim, 2018) may give some guidelines for the NLP community broadly, and work on richer formal systems of guarantees on fairness is underway (Kearns et al., 2019); but while this research is ongoing, developers continue to build systems.”
Bayesian Modeling of Intersectional Fairness: The Variance of Bias (2018, background): “2 Subgroup Fairness [17] proposed multiattribute fairness definitions which aim to prevent fairness gerrymandering at the intersections of protected groups, as later empirically validated by [18].”
A Survey on the Fairness of Recommender Systems (2022, background): “Fairness should be considered in these subgroups, as even if the groups under each single attribute division are fair, subgroups may be unfair to each other [46, 47].”
Towards causal benchmarking of bias in face analysis algorithms (2020, background): “” One simple instance of this problem is sample bias: samples in the wild may fail to represent specific combinations of variables of interest [38, 39, 40].”
A survey on datasets for fairness‐aware machine learning (2021, background): “Especially, in the research (Kearns et al., 2019), they consider atttributes romatic (relationship) and dalc, walc (alcohol consumption) as the protected attributes.”
Themis: automatically testing software for discrimination (2018, background): “The group discrimination definition can be generalized to rich subgroups [30, 30].”