Repairing without Retraining: Avoiding Disparate Impact with Counterfactual Distributions

By Hao Wang, Berk Ustun, Flávio du Pin Calmon. 2019

97 citations recorded by the source corpus.

When the performance of a machine learning model varies over groups defined by sensitive attributes (e.g., gender or ethnicity), the performance disparity can be expressed in terms of the probability distributions of the input and output variables over each group. In this paper, we exploit this fact to reduce the disparate impact of a fixed classification model over a population of interest. Given a black-box classifier, we aim to eliminate the performance gap by perturbing the distribution of input variables for the disadvantaged group. We refer to the perturbed distribution as a counterfactual distribution, and characterize its properties for common fairness criteria. We introduce a descent algorithm to learn a counterfactual distribution from data. We then discuss how the estimated distribution can be used to build a data preprocessor that can reduce disparate impact without training a new model. We validate our approach through experiments on real-world datasets, showing that it can repair different forms of disparity without a significant drop in accuracy.

Subfields: Artificial Intelligence, 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 Artificial Intelligence

In Safety Research

Citations in the corpus

Learning Optimized Risk Scores (2016, neutral): “[115] Wang, Hao, Berk Ustun, and Flavio Calmon.”
Fairness in Machine Learning: A Survey (2020, unknown): “1, we note [273] as an exception here who omit proxies), which can result in increasing instead of decreasing bias and discrimination [170], or indirectly concealing discrimination [86].”
Fairness without Harm: Decoupled Classifiers with Preference Guarantees (2019, background): “In addition to Zafar et al. (2017b) and Dwork et al. (2018), several other works discuss treatment disparity to achieve parity-based notions of fairness (see e.g., Corbett-Davies & Goel, 2018; Kleinberg et al., 2018; Lipton et al., 2018; Wang et al., 2019).”
On the Accuracy of Influence Functions for Measuring Group Effects (2019, background): “…have been recently applied to explain predictions (Koh and Liang, 2017), produce conﬁdence intervals (Schulam and Saria, 2019), investigate model bias (Brunet et al., 2018; Wang et al., 2019), improve human trust (Zhou et al., 2019), and even craft data poisoning attacks (Koh et al., 2019).”
A General Approach to Fairness with Optimal Transport (2020, methodology): “With the exception of Jiang et al. (2019) and Risser et al. (2019), optimal transport has been used to transport model inputs and, with the exception of Wang, Ustun, and Calmon (2019) and Black, Yeom, and Fredrikson (2019), in order to achieve strong demographic parity.”